Currently, uncompressed data except for tail-packing inline is
consecutive on disk.
In order to support chunk-based data deduplication, add a new
corresponding inode data layout.
In the future, the data source of chunks can be either (un)compressed.
Signed-off-by: Gao Xiang <[email protected]>
---
Documentation/filesystems/erofs.rst | 16 ++++++++++--
fs/erofs/erofs_fs.h | 40 +++++++++++++++++++++++++++--
2 files changed, 52 insertions(+), 4 deletions(-)
diff --git a/Documentation/filesystems/erofs.rst b/Documentation/filesystems/erofs.rst
index 868e3972227f..b46d0fc46eb6 100644
--- a/Documentation/filesystems/erofs.rst
+++ b/Documentation/filesystems/erofs.rst
@@ -156,13 +156,14 @@ may not. All metadatas can be now observed in two different spaces (views):
Xattrs, extents, data inline are followed by the corresponding inode with
proper alignment, and they could be optional for different data mappings.
- _currently_ total 4 valid data mappings are supported:
+ _currently_ total 5 data layouts are supported:
== ====================================================================
0 flat file data without data inline (no extent);
1 fixed-sized output data compression (with non-compacted indexes);
2 flat file data with tail packing data inline (no extent);
- 3 fixed-sized output data compression (with compacted indexes, v5.3+).
+ 3 fixed-sized output data compression (with compacted indexes, v5.3+);
+ 4 chunk-based file (v5.15+).
== ====================================================================
The size of the optional xattrs is indicated by i_xattr_count in inode
@@ -213,6 +214,17 @@ Note that apart from the offset of the first filename, nameoff0 also indicates
the total number of directory entries in this block since it is no need to
introduce another on-disk field at all.
+Chunk-based file
+----------------
+In order to support chunk-based file deduplication, a new inode data layout has
+been supported since Linux v5.15: Files are split in equal-sized data chunks
+with ``extents`` area of the inode metadata indicating how to get the chunk
+data: these can be simply as a 4-byte block address array or in the 8-byte
+chunk index form (see struct erofs_inode_chunk_index in erofs_fs.h for more
+details.)
+
+By the way, chunk-based files are all uncompressed for now.
+
Data compression
----------------
EROFS implements LZ4 fixed-sized output compression which generates fixed-sized
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 0f8da74570b4..6210fe434930 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -4,6 +4,7 @@
*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#ifndef __EROFS_FS_H
#define __EROFS_FS_H
@@ -19,10 +20,12 @@
#define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING 0x00000001
#define EROFS_FEATURE_INCOMPAT_COMPR_CFGS 0x00000002
#define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER 0x00000002
+#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE 0x00000004
#define EROFS_ALL_FEATURE_INCOMPAT \
(EROFS_FEATURE_INCOMPAT_LZ4_0PADDING | \
EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
- EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER)
+ EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
+ EROFS_FEATURE_INCOMPAT_CHUNKED_FILE)
#define EROFS_SB_EXTSLOT_SIZE 16
@@ -64,13 +67,16 @@ struct erofs_super_block {
* inode, [xattrs], last_inline_data, ... | ... | no-holed data
* 3 - inode compression D:
* inode, [xattrs], map_header, extents ... | ...
- * 4~7 - reserved
+ * 4 - inode chunk-based E:
+ * inode, [xattrs], chunk indexes ... | ...
+ * 5~7 - reserved
*/
enum {
EROFS_INODE_FLAT_PLAIN = 0,
EROFS_INODE_FLAT_COMPRESSION_LEGACY = 1,
EROFS_INODE_FLAT_INLINE = 2,
EROFS_INODE_FLAT_COMPRESSION = 3,
+ EROFS_INODE_CHUNK_BASED = 4,
EROFS_INODE_DATALAYOUT_MAX
};
@@ -90,6 +96,19 @@ static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
#define EROFS_I_ALL \
((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
+/* indicate chunk blkbits, thus `chunksize = blocksize << chunk blkbits' */
+#define EROFS_CHUNK_FORMAT_BLKBITS_MASK 0x001F
+/* with chunk indexes or just a 4-byte blkaddr array */
+#define EROFS_CHUNK_FORMAT_INDEXES 0x0020
+
+#define EROFS_CHUNK_FORMAT_ALL \
+ (EROFS_CHUNK_FORMAT_BLKBITS_MASK | EROFS_CHUNK_FORMAT_INDEXES)
+
+struct erofs_inode_chunk_info {
+ __le16 format; /* chunk blkbits */
+ __le16 reserved;
+};
+
/* 32-byte reduced form of an ondisk inode */
struct erofs_inode_compact {
__le16 i_format; /* inode format hints */
@@ -107,6 +126,9 @@ struct erofs_inode_compact {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
__le32 i_ino; /* only used for 32-bit stat compatibility */
__le16 i_uid;
@@ -135,6 +157,9 @@ struct erofs_inode_extended {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
/* only used for 32-bit stat compatibility */
@@ -204,6 +229,15 @@ static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
e->e_name_len + le16_to_cpu(e->e_value_size));
}
+/* represent a zeroed chunk (hole) */
+#define EROFS_NULL_ADDR -1
+
+struct erofs_inode_chunk_index {
+ __le32 blkaddr;
+ __le16 device_id; /* back-end storage id, always 0 for now */
+ __le16 reserved; /* reserved, don't care */
+};
+
/* maximum supported size of a physical compression cluster */
#define Z_EROFS_PCLUSTER_MAX_SIZE (1024 * 1024)
@@ -338,6 +372,8 @@ static inline void erofs_check_ondisk_layout_definitions(void)
BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_vle_decompressed_index) != 8);
BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
--
2.24.4
On Wed, Aug 18, 2021 at 03:07:12PM +0800, Gao Xiang wrote:
> Currently, uncompressed data except for tail-packing inline is
> consecutive on disk.
>
> In order to support chunk-based data deduplication, add a new
> corresponding inode data layout.
>
> In the future, the data source of chunks can be either (un)compressed.
>
> Signed-off-by: Gao Xiang <[email protected]>
> ---
> Documentation/filesystems/erofs.rst | 16 ++++++++++--
> fs/erofs/erofs_fs.h | 40 +++++++++++++++++++++++++++--
> 2 files changed, 52 insertions(+), 4 deletions(-)
>
> diff --git a/Documentation/filesystems/erofs.rst b/Documentation/filesystems/erofs.rst
> index 868e3972227f..b46d0fc46eb6 100644
> --- a/Documentation/filesystems/erofs.rst
> +++ b/Documentation/filesystems/erofs.rst
> @@ -156,13 +156,14 @@ may not. All metadatas can be now observed in two different spaces (views):
>
> Xattrs, extents, data inline are followed by the corresponding inode with
> proper alignment, and they could be optional for different data mappings.
> - _currently_ total 4 valid data mappings are supported:
> + _currently_ total 5 data layouts are supported:
>
> == ====================================================================
> 0 flat file data without data inline (no extent);
> 1 fixed-sized output data compression (with non-compacted indexes);
> 2 flat file data with tail packing data inline (no extent);
> - 3 fixed-sized output data compression (with compacted indexes, v5.3+).
> + 3 fixed-sized output data compression (with compacted indexes, v5.3+);
> + 4 chunk-based file (v5.15+).
> == ====================================================================
>
> The size of the optional xattrs is indicated by i_xattr_count in inode
> @@ -213,6 +214,17 @@ Note that apart from the offset of the first filename, nameoff0 also indicates
> the total number of directory entries in this block since it is no need to
> introduce another on-disk field at all.
>
> +Chunk-based file
> +----------------
> +In order to support chunk-based file deduplication, a new inode data layout has
> +been supported since Linux v5.15: Files are split in equal-sized data chunks
> +with ``extents`` area of the inode metadata indicating how to get the chunk
> +data: these can be simply as a 4-byte block address array or in the 8-byte
> +chunk index form (see struct erofs_inode_chunk_index in erofs_fs.h for more
> +details.)
> +
> +By the way, chunk-based files are all uncompressed for now.
> +
> Data compression
> ----------------
> EROFS implements LZ4 fixed-sized output compression which generates fixed-sized
> diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
> index 0f8da74570b4..6210fe434930 100644
> --- a/fs/erofs/erofs_fs.h
> +++ b/fs/erofs/erofs_fs.h
> @@ -4,6 +4,7 @@
> *
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #ifndef __EROFS_FS_H
> #define __EROFS_FS_H
> @@ -19,10 +20,12 @@
> #define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING 0x00000001
> #define EROFS_FEATURE_INCOMPAT_COMPR_CFGS 0x00000002
> #define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER 0x00000002
> +#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE 0x00000004
> #define EROFS_ALL_FEATURE_INCOMPAT \
> (EROFS_FEATURE_INCOMPAT_LZ4_0PADDING | \
> EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
> - EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER)
> + EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
> + EROFS_FEATURE_INCOMPAT_CHUNKED_FILE)
>
> #define EROFS_SB_EXTSLOT_SIZE 16
>
> @@ -64,13 +67,16 @@ struct erofs_super_block {
> * inode, [xattrs], last_inline_data, ... | ... | no-holed data
> * 3 - inode compression D:
> * inode, [xattrs], map_header, extents ... | ...
> - * 4~7 - reserved
> + * 4 - inode chunk-based E:
> + * inode, [xattrs], chunk indexes ... | ...
> + * 5~7 - reserved
> */
> enum {
> EROFS_INODE_FLAT_PLAIN = 0,
> EROFS_INODE_FLAT_COMPRESSION_LEGACY = 1,
> EROFS_INODE_FLAT_INLINE = 2,
> EROFS_INODE_FLAT_COMPRESSION = 3,
> + EROFS_INODE_CHUNK_BASED = 4,
> EROFS_INODE_DATALAYOUT_MAX
> };
>
> @@ -90,6 +96,19 @@ static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
> #define EROFS_I_ALL \
> ((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
>
> +/* indicate chunk blkbits, thus `chunksize = blocksize << chunk blkbits' */
A typo in the quotation marks. (`chunksize = ) should be ('chunksize =)
Otherwise it looks good.
Reviewed-by: Liu Bo <[email protected]>
thanks,
liubo
> +#define EROFS_CHUNK_FORMAT_BLKBITS_MASK 0x001F
> +/* with chunk indexes or just a 4-byte blkaddr array */
> +#define EROFS_CHUNK_FORMAT_INDEXES 0x0020
> +
> +#define EROFS_CHUNK_FORMAT_ALL \
> + (EROFS_CHUNK_FORMAT_BLKBITS_MASK | EROFS_CHUNK_FORMAT_INDEXES)
> +
> +struct erofs_inode_chunk_info {
> + __le16 format; /* chunk blkbits */
> + __le16 reserved;
> +};
> +
> /* 32-byte reduced form of an ondisk inode */
> struct erofs_inode_compact {
> __le16 i_format; /* inode format hints */
> @@ -107,6 +126,9 @@ struct erofs_inode_compact {
>
> /* for device files, used to indicate old/new device # */
> __le32 rdev;
> +
> + /* for chunk-based files, it contains the summary info */
> + struct erofs_inode_chunk_info c;
> } i_u;
> __le32 i_ino; /* only used for 32-bit stat compatibility */
> __le16 i_uid;
> @@ -135,6 +157,9 @@ struct erofs_inode_extended {
>
> /* for device files, used to indicate old/new device # */
> __le32 rdev;
> +
> + /* for chunk-based files, it contains the summary info */
> + struct erofs_inode_chunk_info c;
> } i_u;
>
> /* only used for 32-bit stat compatibility */
> @@ -204,6 +229,15 @@ static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
> e->e_name_len + le16_to_cpu(e->e_value_size));
> }
>
> +/* represent a zeroed chunk (hole) */
> +#define EROFS_NULL_ADDR -1
> +
> +struct erofs_inode_chunk_index {
> + __le32 blkaddr;
> + __le16 device_id; /* back-end storage id, always 0 for now */
> + __le16 reserved; /* reserved, don't care */
> +};
> +
> /* maximum supported size of a physical compression cluster */
> #define Z_EROFS_PCLUSTER_MAX_SIZE (1024 * 1024)
>
> @@ -338,6 +372,8 @@ static inline void erofs_check_ondisk_layout_definitions(void)
> BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
> BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
> BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
> + BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
> + BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
> BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
> BUILD_BUG_ON(sizeof(struct z_erofs_vle_decompressed_index) != 8);
> BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
> --
> 2.24.4
Hi Bo,
On Thu, Aug 19, 2021 at 06:28:04AM +0800, Liu Bo wrote:
> On Wed, Aug 18, 2021 at 03:07:12PM +0800, Gao Xiang wrote:
...
> > + EROFS_INODE_CHUNK_BASED = 4,
> > EROFS_INODE_DATALAYOUT_MAX
> > };
> >
> > @@ -90,6 +96,19 @@ static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
> > #define EROFS_I_ALL \
> > ((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
> >
> > +/* indicate chunk blkbits, thus `chunksize = blocksize << chunk blkbits' */
>
> A typo in the quotation marks. (`chunksize = ) should be ('chunksize =)
Such usage is like below:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=0e389028ad75412ff624b304913bba14f8d46ec4
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=78128fabd022240852859c0b253972147593690b
I'm fine in either way. I'll update it in the next version or when
submitting.
>
> Otherwise it looks good.
>
> Reviewed-by: Liu Bo <[email protected]>
Thanks for the review!
Thanks,
Gao Xiang
>
> thanks,
> liubo
>
On 2021/8/18 15:07, Gao Xiang wrote:
> Currently, uncompressed data except for tail-packing inline is
> consecutive on disk.
>
> In order to support chunk-based data deduplication, add a new
> corresponding inode data layout.
>
> In the future, the data source of chunks can be either (un)compressed.
>
> Signed-off-by: Gao Xiang <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Thanks,
Currently, uncompressed data except for tail-packing inline is
consecutive on disk.
In order to support chunk-based data deduplication, add a new
corresponding inode data layout.
In the future, the data source of chunks can be either (un)compressed.
Reviewed-by: Liu Bo <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Signed-off-by: Gao Xiang <[email protected]>
---
changes since v1:
- update a quotation mark pair suggested by Bo;
- adjust the members of struct erofs_inode_chunk_index to keep in
sync with struct z_erofs_vle_decompressed_index for better
extendibility later.
Documentation/filesystems/erofs.rst | 16 +++++++++--
fs/erofs/erofs_fs.h | 43 +++++++++++++++++++++++++++--
2 files changed, 55 insertions(+), 4 deletions(-)
diff --git a/Documentation/filesystems/erofs.rst b/Documentation/filesystems/erofs.rst
index 868e3972227f..d484408a90c1 100644
--- a/Documentation/filesystems/erofs.rst
+++ b/Documentation/filesystems/erofs.rst
@@ -156,13 +156,14 @@ may not. All metadatas can be now observed in two different spaces (views):
Xattrs, extents, data inline are followed by the corresponding inode with
proper alignment, and they could be optional for different data mappings.
- _currently_ total 4 valid data mappings are supported:
+ _currently_ total 5 data layouts are supported:
== ====================================================================
0 flat file data without data inline (no extent);
1 fixed-sized output data compression (with non-compacted indexes);
2 flat file data with tail packing data inline (no extent);
- 3 fixed-sized output data compression (with compacted indexes, v5.3+).
+ 3 fixed-sized output data compression (with compacted indexes, v5.3+);
+ 4 chunk-based file (v5.15+).
== ====================================================================
The size of the optional xattrs is indicated by i_xattr_count in inode
@@ -213,6 +214,17 @@ Note that apart from the offset of the first filename, nameoff0 also indicates
the total number of directory entries in this block since it is no need to
introduce another on-disk field at all.
+Chunk-based file
+----------------
+In order to support chunk-based file deduplication, a new inode data layout has
+been supported since Linux v5.15: Files are split in equal-sized data chunks
+with ``extents`` area of the inode metadata indicating how to get the chunk
+data: these can be simply as a 4-byte block address array or in the 8-byte
+chunk index form (see struct erofs_inode_chunk_index in erofs_fs.h for more
+details.)
+
+By the way, chunk-based files are all uncompressed for now.
+
Data compression
----------------
EROFS implements LZ4 fixed-sized output compression which generates fixed-sized
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 0f8da74570b4..f4c249996341 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -4,6 +4,7 @@
*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#ifndef __EROFS_FS_H
#define __EROFS_FS_H
@@ -19,10 +20,12 @@
#define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING 0x00000001
#define EROFS_FEATURE_INCOMPAT_COMPR_CFGS 0x00000002
#define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER 0x00000002
+#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE 0x00000004
#define EROFS_ALL_FEATURE_INCOMPAT \
(EROFS_FEATURE_INCOMPAT_LZ4_0PADDING | \
EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
- EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER)
+ EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
+ EROFS_FEATURE_INCOMPAT_CHUNKED_FILE)
#define EROFS_SB_EXTSLOT_SIZE 16
@@ -64,13 +67,16 @@ struct erofs_super_block {
* inode, [xattrs], last_inline_data, ... | ... | no-holed data
* 3 - inode compression D:
* inode, [xattrs], map_header, extents ... | ...
- * 4~7 - reserved
+ * 4 - inode chunk-based E:
+ * inode, [xattrs], chunk indexes ... | ...
+ * 5~7 - reserved
*/
enum {
EROFS_INODE_FLAT_PLAIN = 0,
EROFS_INODE_FLAT_COMPRESSION_LEGACY = 1,
EROFS_INODE_FLAT_INLINE = 2,
EROFS_INODE_FLAT_COMPRESSION = 3,
+ EROFS_INODE_CHUNK_BASED = 4,
EROFS_INODE_DATALAYOUT_MAX
};
@@ -90,6 +96,19 @@ static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
#define EROFS_I_ALL \
((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
+/* indicate chunk blkbits, thus 'chunksize = blocksize << chunk blkbits' */
+#define EROFS_CHUNK_FORMAT_BLKBITS_MASK 0x001F
+/* with chunk indexes or just a 4-byte blkaddr array */
+#define EROFS_CHUNK_FORMAT_INDEXES 0x0020
+
+#define EROFS_CHUNK_FORMAT_ALL \
+ (EROFS_CHUNK_FORMAT_BLKBITS_MASK | EROFS_CHUNK_FORMAT_INDEXES)
+
+struct erofs_inode_chunk_info {
+ __le16 format; /* chunk blkbits, etc. */
+ __le16 reserved;
+};
+
/* 32-byte reduced form of an ondisk inode */
struct erofs_inode_compact {
__le16 i_format; /* inode format hints */
@@ -107,6 +126,9 @@ struct erofs_inode_compact {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
__le32 i_ino; /* only used for 32-bit stat compatibility */
__le16 i_uid;
@@ -135,6 +157,9 @@ struct erofs_inode_extended {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
/* only used for 32-bit stat compatibility */
@@ -204,6 +229,15 @@ static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
e->e_name_len + le16_to_cpu(e->e_value_size));
}
+/* represent a zeroed chunk (hole) */
+#define EROFS_NULL_ADDR -1
+
+struct erofs_inode_chunk_index {
+ __le16 advise; /* always 0, don't care for now */
+ __le16 device_id; /* back-end storage id, always 0 for now */
+ __le32 blkaddr; /* start block address of this inode chunk */
+};
+
/* maximum supported size of a physical compression cluster */
#define Z_EROFS_PCLUSTER_MAX_SIZE (1024 * 1024)
@@ -338,9 +372,14 @@ static inline void erofs_check_ondisk_layout_definitions(void)
BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_vle_decompressed_index) != 8);
BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
+ /* keep in sync between 2 index structures for better extendibility */
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) !=
+ sizeof(struct z_erofs_vle_decompressed_index));
BUILD_BUG_ON(BIT(Z_EROFS_VLE_DI_CLUSTER_TYPE_BITS) <
Z_EROFS_VLE_CLUSTER_TYPE_MAX - 1);
--
2.24.4
Add runtime support for chunk-based uncompressed files
described in the previous patch.
Reviewed-by: Liu Bo <[email protected]>
Signed-off-by: Gao Xiang <[email protected]>
---
changes since v1:
- use le16_to_cpu instead of __le16_to_cpu pointed out by Chao.
fs/erofs/data.c | 90 ++++++++++++++++++++++++++++++++++++++++-----
fs/erofs/inode.c | 18 ++++++++-
fs/erofs/internal.h | 5 +++
3 files changed, 102 insertions(+), 11 deletions(-)
diff --git a/fs/erofs/data.c b/fs/erofs/data.c
index 09c46fbdb9b2..ee9a33485313 100644
--- a/fs/erofs/data.c
+++ b/fs/erofs/data.c
@@ -2,6 +2,7 @@
/*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#include "internal.h"
#include <linux/prefetch.h>
@@ -36,13 +37,6 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
nblocks = DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
lastblk = nblocks - tailendpacking;
- if (offset >= inode->i_size) {
- /* leave out-of-bound access unmapped */
- map->m_flags = 0;
- map->m_plen = 0;
- goto out;
- }
-
/* there is no hole in flatmode */
map->m_flags = EROFS_MAP_MAPPED;
@@ -77,14 +71,90 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
goto err_out;
}
-out:
map->m_llen = map->m_plen;
-
err_out:
trace_erofs_map_blocks_flatmode_exit(inode, map, flags, 0);
return err;
}
+static int erofs_map_blocks(struct inode *inode,
+ struct erofs_map_blocks *map, int flags)
+{
+ struct super_block *sb = inode->i_sb;
+ struct erofs_inode *vi = EROFS_I(inode);
+ struct erofs_inode_chunk_index *idx;
+ struct page *page;
+ u64 chunknr;
+ unsigned int unit;
+ erofs_off_t pos;
+ int err = 0;
+
+ if (map->m_la >= inode->i_size) {
+ /* leave out-of-bound access unmapped */
+ map->m_flags = 0;
+ map->m_plen = 0;
+ goto out;
+ }
+
+ if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
+ return erofs_map_blocks_flatmode(inode, map, flags);
+
+ if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
+ unit = sizeof(*idx); /* chunk index */
+ else
+ unit = 4; /* block map */
+
+ chunknr = map->m_la >> vi->chunkbits;
+ pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize +
+ vi->xattr_isize, unit) + unit * chunknr;
+
+ page = erofs_get_meta_page(inode->i_sb, erofs_blknr(pos));
+ if (IS_ERR(page))
+ return PTR_ERR(page);
+
+ map->m_la = chunknr << vi->chunkbits;
+ map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
+ roundup(inode->i_size - map->m_la, EROFS_BLKSIZ));
+
+ /* handle block map */
+ if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
+ __le32 *blkaddr = page_address(page) + erofs_blkoff(pos);
+
+ if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
+ map->m_flags = 0;
+ } else {
+ map->m_pa = blknr_to_addr(le32_to_cpu(*blkaddr));
+ map->m_flags = EROFS_MAP_MAPPED;
+ }
+ goto out_unlock;
+ }
+ /* parse chunk indexes */
+ idx = page_address(page) + erofs_blkoff(pos);
+ switch (le32_to_cpu(idx->blkaddr)) {
+ case EROFS_NULL_ADDR:
+ map->m_flags = 0;
+ break;
+ default:
+ /* only one device is supported for now */
+ if (idx->device_id) {
+ erofs_err(sb, "invalid device id %u @ %llu for nid %llu",
+ le32_to_cpu(idx->device_id),
+ chunknr, vi->nid);
+ err = -EFSCORRUPTED;
+ goto out_unlock;
+ }
+ map->m_pa = blknr_to_addr(le32_to_cpu(idx->blkaddr));
+ map->m_flags = EROFS_MAP_MAPPED;
+ break;
+ }
+out_unlock:
+ unlock_page(page);
+ put_page(page);
+out:
+ map->m_llen = map->m_plen;
+ return err;
+}
+
static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
{
@@ -94,7 +164,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
map.m_la = offset;
map.m_llen = length;
- ret = erofs_map_blocks_flatmode(inode, &map, EROFS_GET_BLOCKS_RAW);
+ ret = erofs_map_blocks(inode, &map, EROFS_GET_BLOCKS_RAW);
if (ret < 0)
return ret;
diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
index d13e0709599c..4408929bd6f5 100644
--- a/fs/erofs/inode.c
+++ b/fs/erofs/inode.c
@@ -2,6 +2,7 @@
/*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#include "xattr.h"
@@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
/* total blocks for compressed files */
if (erofs_inode_is_data_compressed(vi->datalayout))
nblks = le32_to_cpu(die->i_u.compressed_blocks);
-
+ else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
+ /* fill chunked inode summary info */
+ vi->chunkformat = le16_to_cpu(die->i_u.c.format);
kfree(copied);
break;
case EROFS_INODE_LAYOUT_COMPACT:
@@ -160,6 +163,8 @@ static struct page *erofs_read_inode(struct inode *inode,
inode->i_size = le32_to_cpu(dic->i_size);
if (erofs_inode_is_data_compressed(vi->datalayout))
nblks = le32_to_cpu(dic->i_u.compressed_blocks);
+ else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
+ vi->chunkformat = le16_to_cpu(dic->i_u.c.format);
break;
default:
erofs_err(inode->i_sb,
@@ -169,6 +174,17 @@ static struct page *erofs_read_inode(struct inode *inode,
goto err_out;
}
+ if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
+ if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_ALL)) {
+ erofs_err(inode->i_sb,
+ "unsupported chunk format %x of nid %llu",
+ vi->chunkformat, vi->nid);
+ err = -EOPNOTSUPP;
+ goto err_out;
+ }
+ vi->chunkbits = LOG_BLOCK_SIZE +
+ (vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
+ }
inode->i_mtime.tv_sec = inode->i_ctime.tv_sec;
inode->i_atime.tv_sec = inode->i_ctime.tv_sec;
inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
index 91089ab8a816..9524e155b38f 100644
--- a/fs/erofs/internal.h
+++ b/fs/erofs/internal.h
@@ -2,6 +2,7 @@
/*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#ifndef __EROFS_INTERNAL_H
#define __EROFS_INTERNAL_H
@@ -261,6 +262,10 @@ struct erofs_inode {
union {
erofs_blk_t raw_blkaddr;
+ struct {
+ unsigned short chunkformat;
+ unsigned char chunkbits;
+ };
#ifdef CONFIG_EROFS_FS_ZIP
struct {
unsigned short z_advise;
--
2.24.4
On 8/19/21 2:33 PM, Gao Xiang wrote:
> Add runtime support for chunk-based uncompressed files
> described in the previous patch.
>
> Reviewed-by: Liu Bo <[email protected]>
> Signed-off-by: Gao Xiang <[email protected]>
> ---
> changes since v1:
> - use le16_to_cpu instead of __le16_to_cpu pointed out by Chao.
>
> fs/erofs/data.c | 90 ++++++++++++++++++++++++++++++++++++++++-----
> fs/erofs/inode.c | 18 ++++++++-
> fs/erofs/internal.h | 5 +++
> 3 files changed, 102 insertions(+), 11 deletions(-)
>
> diff --git a/fs/erofs/data.c b/fs/erofs/data.c
> index 09c46fbdb9b2..ee9a33485313 100644
> --- a/fs/erofs/data.c
> +++ b/fs/erofs/data.c
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #include "internal.h"
> #include <linux/prefetch.h>
> @@ -36,13 +37,6 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
> nblocks = DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
> lastblk = nblocks - tailendpacking;
>
> - if (offset >= inode->i_size) {
> - /* leave out-of-bound access unmapped */
> - map->m_flags = 0;
> - map->m_plen = 0;
> - goto out;
> - }
> -
> /* there is no hole in flatmode */
> map->m_flags = EROFS_MAP_MAPPED;
>
> @@ -77,14 +71,90 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
> goto err_out;
> }
>
> -out:
> map->m_llen = map->m_plen;
> -
> err_out:
> trace_erofs_map_blocks_flatmode_exit(inode, map, flags, 0);
> return err;
> }
>
> +static int erofs_map_blocks(struct inode *inode,
> + struct erofs_map_blocks *map, int flags)
> +{
> + struct super_block *sb = inode->i_sb;
> + struct erofs_inode *vi = EROFS_I(inode);
> + struct erofs_inode_chunk_index *idx;
> + struct page *page;
> + u64 chunknr;
> + unsigned int unit;
> + erofs_off_t pos;
> + int err = 0;
> +
> + if (map->m_la >= inode->i_size) {
> + /* leave out-of-bound access unmapped */
> + map->m_flags = 0;
> + map->m_plen = 0;
> + goto out;
> + }
> +
> + if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
> + return erofs_map_blocks_flatmode(inode, map, flags);
> +
> + if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
> + unit = sizeof(*idx); /* chunk index */
> + else
> + unit = 4; /* block map */
> +
> + chunknr = map->m_la >> vi->chunkbits;
> + pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize +
> + vi->xattr_isize, unit) + unit * chunknr;
> +
> + page = erofs_get_meta_page(inode->i_sb, erofs_blknr(pos));
> + if (IS_ERR(page))
> + return PTR_ERR(page);
> +
> + map->m_la = chunknr << vi->chunkbits;
> + map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
> + roundup(inode->i_size - map->m_la, EROFS_BLKSIZ));
> +
> + /* handle block map */
> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
> + __le32 *blkaddr = page_address(page) + erofs_blkoff(pos);
> +
> + if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
> + map->m_flags = 0;
> + } else {
> + map->m_pa = blknr_to_addr(le32_to_cpu(*blkaddr));
> + map->m_flags = EROFS_MAP_MAPPED;
> + }
> + goto out_unlock;
> + }
> + /* parse chunk indexes */
> + idx = page_address(page) + erofs_blkoff(pos);
> + switch (le32_to_cpu(idx->blkaddr)) {
> + case EROFS_NULL_ADDR:
> + map->m_flags = 0;
> + break;
> + default:
> + /* only one device is supported for now */
> + if (idx->device_id) {
> + erofs_err(sb, "invalid device id %u @ %llu for nid %llu",
> + le32_to_cpu(idx->device_id),
> + chunknr, vi->nid);
> + err = -EFSCORRUPTED;
> + goto out_unlock;
> + }
> + map->m_pa = blknr_to_addr(le32_to_cpu(idx->blkaddr));
> + map->m_flags = EROFS_MAP_MAPPED;
> + break;
> + }
> +out_unlock:
> + unlock_page(page);
> + put_page(page);
> +out:
> + map->m_llen = map->m_plen;
> + return err;
> +}
> +
> static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
> {
> @@ -94,7 +164,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> map.m_la = offset;
> map.m_llen = length;
>
> - ret = erofs_map_blocks_flatmode(inode, &map, EROFS_GET_BLOCKS_RAW);
> + ret = erofs_map_blocks(inode, &map, EROFS_GET_BLOCKS_RAW);
> if (ret < 0)
> return ret;
>
> diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
> index d13e0709599c..4408929bd6f5 100644
> --- a/fs/erofs/inode.c
> +++ b/fs/erofs/inode.c
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #include "xattr.h"
>
> @@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
> /* total blocks for compressed files */
> if (erofs_inode_is_data_compressed(vi->datalayout))
> nblks = le32_to_cpu(die->i_u.compressed_blocks);
> -
> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> + /* fill chunked inode summary info */
> + vi->chunkformat = le16_to_cpu(die->i_u.c.format);
Better to add braces for if/else.
Thanks,
Joseph
> kfree(copied);
> break;
> case EROFS_INODE_LAYOUT_COMPACT:
> @@ -160,6 +163,8 @@ static struct page *erofs_read_inode(struct inode *inode,
> inode->i_size = le32_to_cpu(dic->i_size);
> if (erofs_inode_is_data_compressed(vi->datalayout))
> nblks = le32_to_cpu(dic->i_u.compressed_blocks);
> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> + vi->chunkformat = le16_to_cpu(dic->i_u.c.format);
> break;
> default:
> erofs_err(inode->i_sb,
> @@ -169,6 +174,17 @@ static struct page *erofs_read_inode(struct inode *inode,
> goto err_out;
> }
>
> + if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_ALL)) {
> + erofs_err(inode->i_sb,
> + "unsupported chunk format %x of nid %llu",
> + vi->chunkformat, vi->nid);
> + err = -EOPNOTSUPP;
> + goto err_out;
> + }
> + vi->chunkbits = LOG_BLOCK_SIZE +
> + (vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
> + }
> inode->i_mtime.tv_sec = inode->i_ctime.tv_sec;
> inode->i_atime.tv_sec = inode->i_ctime.tv_sec;
> inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
> diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
> index 91089ab8a816..9524e155b38f 100644
> --- a/fs/erofs/internal.h
> +++ b/fs/erofs/internal.h
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #ifndef __EROFS_INTERNAL_H
> #define __EROFS_INTERNAL_H
> @@ -261,6 +262,10 @@ struct erofs_inode {
>
> union {
> erofs_blk_t raw_blkaddr;
> + struct {
> + unsigned short chunkformat;
> + unsigned char chunkbits;
> + };
> #ifdef CONFIG_EROFS_FS_ZIP
> struct {
> unsigned short z_advise;
>
Hi Joseph,
On Thu, Aug 19, 2021 at 02:37:50PM +0800, Joseph Qi wrote:
>
>
> On 8/19/21 2:33 PM, Gao Xiang wrote:
...
> > diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
> > index d13e0709599c..4408929bd6f5 100644
> > --- a/fs/erofs/inode.c
> > +++ b/fs/erofs/inode.c
> > @@ -2,6 +2,7 @@
> > /*
> > * Copyright (C) 2017-2018 HUAWEI, Inc.
> > * https://www.huawei.com/
> > + * Copyright (C) 2021, Alibaba Cloud
> > */
> > #include "xattr.h"
> >
> > @@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
> > /* total blocks for compressed files */
> > if (erofs_inode_is_data_compressed(vi->datalayout))
> > nblks = le32_to_cpu(die->i_u.compressed_blocks);
> > -
> > + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> > + /* fill chunked inode summary info */
> > + vi->chunkformat = le16_to_cpu(die->i_u.c.format);
>
> Better to add braces for if/else.
Thanks for the kind suggestion. Here is single statement, I've checked
coding-style in Documentation. It's no necessary to use brace for this.
And checkpatch didn't report anything.
Also, I found some reference at
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/vmscan.c?h=v5.13#n3066
But anyway, I could update it when applying, either looks good to me.
Thanks,
Gao Xiang
On 2021/8/19 14:33, Gao Xiang wrote:
> Add runtime support for chunk-based uncompressed files
> described in the previous patch.
>
> Reviewed-by: Liu Bo <[email protected]>
> Signed-off-by: Gao Xiang <[email protected]>
> ---
> changes since v1:
> - use le16_to_cpu instead of __le16_to_cpu pointed out by Chao.
>
> fs/erofs/data.c | 90 ++++++++++++++++++++++++++++++++++++++++-----
> fs/erofs/inode.c | 18 ++++++++-
> fs/erofs/internal.h | 5 +++
> 3 files changed, 102 insertions(+), 11 deletions(-)
>
> diff --git a/fs/erofs/data.c b/fs/erofs/data.c
> index 09c46fbdb9b2..ee9a33485313 100644
> --- a/fs/erofs/data.c
> +++ b/fs/erofs/data.c
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #include "internal.h"
> #include <linux/prefetch.h>
> @@ -36,13 +37,6 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
> nblocks = DIV_ROUND_UP(inode->i_size, PAGE_SIZE);
> lastblk = nblocks - tailendpacking;
>
> - if (offset >= inode->i_size) {
> - /* leave out-of-bound access unmapped */
> - map->m_flags = 0;
> - map->m_plen = 0;
> - goto out;
> - }
> -
> /* there is no hole in flatmode */
> map->m_flags = EROFS_MAP_MAPPED;
>
> @@ -77,14 +71,90 @@ static int erofs_map_blocks_flatmode(struct inode *inode,
> goto err_out;
> }
>
> -out:
> map->m_llen = map->m_plen;
> -
> err_out:
> trace_erofs_map_blocks_flatmode_exit(inode, map, flags, 0);
> return err;
> }
>
> +static int erofs_map_blocks(struct inode *inode,
> + struct erofs_map_blocks *map, int flags)
> +{
> + struct super_block *sb = inode->i_sb;
> + struct erofs_inode *vi = EROFS_I(inode);
> + struct erofs_inode_chunk_index *idx;
> + struct page *page;
> + u64 chunknr;
> + unsigned int unit;
> + erofs_off_t pos;
> + int err = 0;
> +
> + if (map->m_la >= inode->i_size) {
> + /* leave out-of-bound access unmapped */
> + map->m_flags = 0;
> + map->m_plen = 0;
> + goto out;
> + }
> +
> + if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
> + return erofs_map_blocks_flatmode(inode, map, flags);
> +
> + if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
> + unit = sizeof(*idx); /* chunk index */
> + else
> + unit = 4; /* block map */
You mean sizeof(__le32)?
Otherwise it looks good to me.
Reviewed-by: Chao Yu <[email protected]>
Thanks,
> +
> + chunknr = map->m_la >> vi->chunkbits;
> + pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize +
> + vi->xattr_isize, unit) + unit * chunknr;
> +
> + page = erofs_get_meta_page(inode->i_sb, erofs_blknr(pos));
> + if (IS_ERR(page))
> + return PTR_ERR(page);
> +
> + map->m_la = chunknr << vi->chunkbits;
> + map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
> + roundup(inode->i_size - map->m_la, EROFS_BLKSIZ));
> +
> + /* handle block map */
> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
> + __le32 *blkaddr = page_address(page) + erofs_blkoff(pos);
> +
> + if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
> + map->m_flags = 0;
> + } else {
> + map->m_pa = blknr_to_addr(le32_to_cpu(*blkaddr));
> + map->m_flags = EROFS_MAP_MAPPED;
> + }
> + goto out_unlock;
> + }
> + /* parse chunk indexes */
> + idx = page_address(page) + erofs_blkoff(pos);
> + switch (le32_to_cpu(idx->blkaddr)) {
> + case EROFS_NULL_ADDR:
> + map->m_flags = 0;
> + break;
> + default:
> + /* only one device is supported for now */
> + if (idx->device_id) {
> + erofs_err(sb, "invalid device id %u @ %llu for nid %llu",
> + le32_to_cpu(idx->device_id),
> + chunknr, vi->nid);
> + err = -EFSCORRUPTED;
> + goto out_unlock;
> + }
> + map->m_pa = blknr_to_addr(le32_to_cpu(idx->blkaddr));
> + map->m_flags = EROFS_MAP_MAPPED;
> + break;
> + }
> +out_unlock:
> + unlock_page(page);
> + put_page(page);
> +out:
> + map->m_llen = map->m_plen;
> + return err;
> +}
> +
> static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
> {
> @@ -94,7 +164,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> map.m_la = offset;
> map.m_llen = length;
>
> - ret = erofs_map_blocks_flatmode(inode, &map, EROFS_GET_BLOCKS_RAW);
> + ret = erofs_map_blocks(inode, &map, EROFS_GET_BLOCKS_RAW);
> if (ret < 0)
> return ret;
>
> diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
> index d13e0709599c..4408929bd6f5 100644
> --- a/fs/erofs/inode.c
> +++ b/fs/erofs/inode.c
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #include "xattr.h"
>
> @@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
> /* total blocks for compressed files */
> if (erofs_inode_is_data_compressed(vi->datalayout))
> nblks = le32_to_cpu(die->i_u.compressed_blocks);
> -
> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> + /* fill chunked inode summary info */
> + vi->chunkformat = le16_to_cpu(die->i_u.c.format);
> kfree(copied);
> break;
> case EROFS_INODE_LAYOUT_COMPACT:
> @@ -160,6 +163,8 @@ static struct page *erofs_read_inode(struct inode *inode,
> inode->i_size = le32_to_cpu(dic->i_size);
> if (erofs_inode_is_data_compressed(vi->datalayout))
> nblks = le32_to_cpu(dic->i_u.compressed_blocks);
> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> + vi->chunkformat = le16_to_cpu(dic->i_u.c.format);
> break;
> default:
> erofs_err(inode->i_sb,
> @@ -169,6 +174,17 @@ static struct page *erofs_read_inode(struct inode *inode,
> goto err_out;
> }
>
> + if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_ALL)) {
> + erofs_err(inode->i_sb,
> + "unsupported chunk format %x of nid %llu",
> + vi->chunkformat, vi->nid);
> + err = -EOPNOTSUPP;
> + goto err_out;
> + }
> + vi->chunkbits = LOG_BLOCK_SIZE +
> + (vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
> + }
> inode->i_mtime.tv_sec = inode->i_ctime.tv_sec;
> inode->i_atime.tv_sec = inode->i_ctime.tv_sec;
> inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
> diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
> index 91089ab8a816..9524e155b38f 100644
> --- a/fs/erofs/internal.h
> +++ b/fs/erofs/internal.h
> @@ -2,6 +2,7 @@
> /*
> * Copyright (C) 2017-2018 HUAWEI, Inc.
> * https://www.huawei.com/
> + * Copyright (C) 2021, Alibaba Cloud
> */
> #ifndef __EROFS_INTERNAL_H
> #define __EROFS_INTERNAL_H
> @@ -261,6 +262,10 @@ struct erofs_inode {
>
> union {
> erofs_blk_t raw_blkaddr;
> + struct {
> + unsigned short chunkformat;
> + unsigned char chunkbits;
> + };
> #ifdef CONFIG_EROFS_FS_ZIP
> struct {
> unsigned short z_advise;
>
Hi Chao,
On Fri, Aug 20, 2021 at 05:04:13PM +0800, Chao Yu wrote:
> On 2021/8/19 14:33, Gao Xiang wrote:
...
> > }
> > +static int erofs_map_blocks(struct inode *inode,
> > + struct erofs_map_blocks *map, int flags)
> > +{
> > + struct super_block *sb = inode->i_sb;
> > + struct erofs_inode *vi = EROFS_I(inode);
> > + struct erofs_inode_chunk_index *idx;
> > + struct page *page;
> > + u64 chunknr;
> > + unsigned int unit;
> > + erofs_off_t pos;
> > + int err = 0;
> > +
> > + if (map->m_la >= inode->i_size) {
> > + /* leave out-of-bound access unmapped */
> > + map->m_flags = 0;
> > + map->m_plen = 0;
> > + goto out;
> > + }
> > +
> > + if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
> > + return erofs_map_blocks_flatmode(inode, map, flags);
> > +
> > + if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
> > + unit = sizeof(*idx); /* chunk index */
> > + else
> > + unit = 4; /* block map */
>
> You mean sizeof(__le32)?
Yeah, sizeof(__le32) == 4, either way works for me.
If some tendency about this, I will update when applying.
>
> Otherwise it looks good to me.
>
> Reviewed-by: Chao Yu <[email protected]>
>
Thanks for the review!
Thanks,
Gao Xiang
> Thanks,
>
> > +
> > + chunknr = map->m_la >> vi->chunkbits;
> > + pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize +
> > + vi->xattr_isize, unit) + unit * chunknr;
> > +
> > + page = erofs_get_meta_page(inode->i_sb, erofs_blknr(pos));
> > + if (IS_ERR(page))
> > + return PTR_ERR(page);
> > +
> > + map->m_la = chunknr << vi->chunkbits;
> > + map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
> > + roundup(inode->i_size - map->m_la, EROFS_BLKSIZ));
> > +
> > + /* handle block map */
> > + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
> > + __le32 *blkaddr = page_address(page) + erofs_blkoff(pos);
> > +
> > + if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
> > + map->m_flags = 0;
> > + } else {
> > + map->m_pa = blknr_to_addr(le32_to_cpu(*blkaddr));
> > + map->m_flags = EROFS_MAP_MAPPED;
> > + }
> > + goto out_unlock;
> > + }
> > + /* parse chunk indexes */
> > + idx = page_address(page) + erofs_blkoff(pos);
> > + switch (le32_to_cpu(idx->blkaddr)) {
> > + case EROFS_NULL_ADDR:
> > + map->m_flags = 0;
> > + break;
> > + default:
> > + /* only one device is supported for now */
> > + if (idx->device_id) {
> > + erofs_err(sb, "invalid device id %u @ %llu for nid %llu",
> > + le32_to_cpu(idx->device_id),
> > + chunknr, vi->nid);
> > + err = -EFSCORRUPTED;
> > + goto out_unlock;
> > + }
> > + map->m_pa = blknr_to_addr(le32_to_cpu(idx->blkaddr));
> > + map->m_flags = EROFS_MAP_MAPPED;
> > + break;
> > + }
> > +out_unlock:
> > + unlock_page(page);
> > + put_page(page);
> > +out:
> > + map->m_llen = map->m_plen;
> > + return err;
> > +}
> > +
> > static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> > unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
> > {
> > @@ -94,7 +164,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
> > map.m_la = offset;
> > map.m_llen = length;
> > - ret = erofs_map_blocks_flatmode(inode, &map, EROFS_GET_BLOCKS_RAW);
> > + ret = erofs_map_blocks(inode, &map, EROFS_GET_BLOCKS_RAW);
> > if (ret < 0)
> > return ret;
> > diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
> > index d13e0709599c..4408929bd6f5 100644
> > --- a/fs/erofs/inode.c
> > +++ b/fs/erofs/inode.c
> > @@ -2,6 +2,7 @@
> > /*
> > * Copyright (C) 2017-2018 HUAWEI, Inc.
> > * https://www.huawei.com/
> > + * Copyright (C) 2021, Alibaba Cloud
> > */
> > #include "xattr.h"
> > @@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
> > /* total blocks for compressed files */
> > if (erofs_inode_is_data_compressed(vi->datalayout))
> > nblks = le32_to_cpu(die->i_u.compressed_blocks);
> > -
> > + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> > + /* fill chunked inode summary info */
> > + vi->chunkformat = le16_to_cpu(die->i_u.c.format);
> > kfree(copied);
> > break;
> > case EROFS_INODE_LAYOUT_COMPACT:
> > @@ -160,6 +163,8 @@ static struct page *erofs_read_inode(struct inode *inode,
> > inode->i_size = le32_to_cpu(dic->i_size);
> > if (erofs_inode_is_data_compressed(vi->datalayout))
> > nblks = le32_to_cpu(dic->i_u.compressed_blocks);
> > + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
> > + vi->chunkformat = le16_to_cpu(dic->i_u.c.format);
> > break;
> > default:
> > erofs_err(inode->i_sb,
> > @@ -169,6 +174,17 @@ static struct page *erofs_read_inode(struct inode *inode,
> > goto err_out;
> > }
> > + if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
> > + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_ALL)) {
> > + erofs_err(inode->i_sb,
> > + "unsupported chunk format %x of nid %llu",
> > + vi->chunkformat, vi->nid);
> > + err = -EOPNOTSUPP;
> > + goto err_out;
> > + }
> > + vi->chunkbits = LOG_BLOCK_SIZE +
> > + (vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
> > + }
> > inode->i_mtime.tv_sec = inode->i_ctime.tv_sec;
> > inode->i_atime.tv_sec = inode->i_ctime.tv_sec;
> > inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
> > diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
> > index 91089ab8a816..9524e155b38f 100644
> > --- a/fs/erofs/internal.h
> > +++ b/fs/erofs/internal.h
> > @@ -2,6 +2,7 @@
> > /*
> > * Copyright (C) 2017-2018 HUAWEI, Inc.
> > * https://www.huawei.com/
> > + * Copyright (C) 2021, Alibaba Cloud
> > */
> > #ifndef __EROFS_INTERNAL_H
> > #define __EROFS_INTERNAL_H
> > @@ -261,6 +262,10 @@ struct erofs_inode {
> > union {
> > erofs_blk_t raw_blkaddr;
> > + struct {
> > + unsigned short chunkformat;
> > + unsigned char chunkbits;
> > + };
> > #ifdef CONFIG_EROFS_FS_ZIP
> > struct {
> > unsigned short z_advise;
> >
On 2021/8/20 17:12, Gao Xiang wrote:
> Hi Chao,
>
> On Fri, Aug 20, 2021 at 05:04:13PM +0800, Chao Yu wrote:
>> On 2021/8/19 14:33, Gao Xiang wrote:
>
> ...
>
>>> }
>>> +static int erofs_map_blocks(struct inode *inode,
>>> + struct erofs_map_blocks *map, int flags)
>>> +{
>>> + struct super_block *sb = inode->i_sb;
>>> + struct erofs_inode *vi = EROFS_I(inode);
>>> + struct erofs_inode_chunk_index *idx;
>>> + struct page *page;
>>> + u64 chunknr;
>>> + unsigned int unit;
>>> + erofs_off_t pos;
>>> + int err = 0;
>>> +
>>> + if (map->m_la >= inode->i_size) {
>>> + /* leave out-of-bound access unmapped */
>>> + map->m_flags = 0;
>>> + map->m_plen = 0;
>>> + goto out;
>>> + }
>>> +
>>> + if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
>>> + return erofs_map_blocks_flatmode(inode, map, flags);
>>> +
>>> + if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
>>> + unit = sizeof(*idx); /* chunk index */
>>> + else
>>> + unit = 4; /* block map */
>>
>> You mean sizeof(__le32)?
>
> Yeah, sizeof(__le32) == 4, either way works for me.
>
> If some tendency about this, I will update when applying.
Xiang,
Yeah, I preper:
#define EROFS_BLOCK_MAP_ENTRY_SIZE sizeof(__le32)
unit = EROFS_BLOCK_MAP_ENTRY_SIZE;
to improve readablity, but unit = sizeof(__le32) is fine as well.
Thanks,
>
>>
>> Otherwise it looks good to me.
>>
>> Reviewed-by: Chao Yu <[email protected]>
>>
>
> Thanks for the review!
>
> Thanks,
> Gao Xiang
>
>> Thanks,
>>
>>> +
>>> + chunknr = map->m_la >> vi->chunkbits;
>>> + pos = ALIGN(iloc(EROFS_SB(sb), vi->nid) + vi->inode_isize +
>>> + vi->xattr_isize, unit) + unit * chunknr;
>>> +
>>> + page = erofs_get_meta_page(inode->i_sb, erofs_blknr(pos));
>>> + if (IS_ERR(page))
>>> + return PTR_ERR(page);
>>> +
>>> + map->m_la = chunknr << vi->chunkbits;
>>> + map->m_plen = min_t(erofs_off_t, 1UL << vi->chunkbits,
>>> + roundup(inode->i_size - map->m_la, EROFS_BLKSIZ));
>>> +
>>> + /* handle block map */
>>> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)) {
>>> + __le32 *blkaddr = page_address(page) + erofs_blkoff(pos);
>>> +
>>> + if (le32_to_cpu(*blkaddr) == EROFS_NULL_ADDR) {
>>> + map->m_flags = 0;
>>> + } else {
>>> + map->m_pa = blknr_to_addr(le32_to_cpu(*blkaddr));
>>> + map->m_flags = EROFS_MAP_MAPPED;
>>> + }
>>> + goto out_unlock;
>>> + }
>>> + /* parse chunk indexes */
>>> + idx = page_address(page) + erofs_blkoff(pos);
>>> + switch (le32_to_cpu(idx->blkaddr)) {
>>> + case EROFS_NULL_ADDR:
>>> + map->m_flags = 0;
>>> + break;
>>> + default:
>>> + /* only one device is supported for now */
>>> + if (idx->device_id) {
>>> + erofs_err(sb, "invalid device id %u @ %llu for nid %llu",
>>> + le32_to_cpu(idx->device_id),
>>> + chunknr, vi->nid);
>>> + err = -EFSCORRUPTED;
>>> + goto out_unlock;
>>> + }
>>> + map->m_pa = blknr_to_addr(le32_to_cpu(idx->blkaddr));
>>> + map->m_flags = EROFS_MAP_MAPPED;
>>> + break;
>>> + }
>>> +out_unlock:
>>> + unlock_page(page);
>>> + put_page(page);
>>> +out:
>>> + map->m_llen = map->m_plen;
>>> + return err;
>>> +}
>>> +
>>> static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>>> unsigned int flags, struct iomap *iomap, struct iomap *srcmap)
>>> {
>>> @@ -94,7 +164,7 @@ static int erofs_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
>>> map.m_la = offset;
>>> map.m_llen = length;
>>> - ret = erofs_map_blocks_flatmode(inode, &map, EROFS_GET_BLOCKS_RAW);
>>> + ret = erofs_map_blocks(inode, &map, EROFS_GET_BLOCKS_RAW);
>>> if (ret < 0)
>>> return ret;
>>> diff --git a/fs/erofs/inode.c b/fs/erofs/inode.c
>>> index d13e0709599c..4408929bd6f5 100644
>>> --- a/fs/erofs/inode.c
>>> +++ b/fs/erofs/inode.c
>>> @@ -2,6 +2,7 @@
>>> /*
>>> * Copyright (C) 2017-2018 HUAWEI, Inc.
>>> * https://www.huawei.com/
>>> + * Copyright (C) 2021, Alibaba Cloud
>>> */
>>> #include "xattr.h"
>>> @@ -122,7 +123,9 @@ static struct page *erofs_read_inode(struct inode *inode,
>>> /* total blocks for compressed files */
>>> if (erofs_inode_is_data_compressed(vi->datalayout))
>>> nblks = le32_to_cpu(die->i_u.compressed_blocks);
>>> -
>>> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
>>> + /* fill chunked inode summary info */
>>> + vi->chunkformat = le16_to_cpu(die->i_u.c.format);
>>> kfree(copied);
>>> break;
>>> case EROFS_INODE_LAYOUT_COMPACT:
>>> @@ -160,6 +163,8 @@ static struct page *erofs_read_inode(struct inode *inode,
>>> inode->i_size = le32_to_cpu(dic->i_size);
>>> if (erofs_inode_is_data_compressed(vi->datalayout))
>>> nblks = le32_to_cpu(dic->i_u.compressed_blocks);
>>> + else if (vi->datalayout == EROFS_INODE_CHUNK_BASED)
>>> + vi->chunkformat = le16_to_cpu(dic->i_u.c.format);
>>> break;
>>> default:
>>> erofs_err(inode->i_sb,
>>> @@ -169,6 +174,17 @@ static struct page *erofs_read_inode(struct inode *inode,
>>> goto err_out;
>>> }
>>> + if (vi->datalayout == EROFS_INODE_CHUNK_BASED) {
>>> + if (!(vi->chunkformat & EROFS_CHUNK_FORMAT_ALL)) {
>>> + erofs_err(inode->i_sb,
>>> + "unsupported chunk format %x of nid %llu",
>>> + vi->chunkformat, vi->nid);
>>> + err = -EOPNOTSUPP;
>>> + goto err_out;
>>> + }
>>> + vi->chunkbits = LOG_BLOCK_SIZE +
>>> + (vi->chunkformat & EROFS_CHUNK_FORMAT_BLKBITS_MASK);
>>> + }
>>> inode->i_mtime.tv_sec = inode->i_ctime.tv_sec;
>>> inode->i_atime.tv_sec = inode->i_ctime.tv_sec;
>>> inode->i_mtime.tv_nsec = inode->i_ctime.tv_nsec;
>>> diff --git a/fs/erofs/internal.h b/fs/erofs/internal.h
>>> index 91089ab8a816..9524e155b38f 100644
>>> --- a/fs/erofs/internal.h
>>> +++ b/fs/erofs/internal.h
>>> @@ -2,6 +2,7 @@
>>> /*
>>> * Copyright (C) 2017-2018 HUAWEI, Inc.
>>> * https://www.huawei.com/
>>> + * Copyright (C) 2021, Alibaba Cloud
>>> */
>>> #ifndef __EROFS_INTERNAL_H
>>> #define __EROFS_INTERNAL_H
>>> @@ -261,6 +262,10 @@ struct erofs_inode {
>>> union {
>>> erofs_blk_t raw_blkaddr;
>>> + struct {
>>> + unsigned short chunkformat;
>>> + unsigned char chunkbits;
>>> + };
>>> #ifdef CONFIG_EROFS_FS_ZIP
>>> struct {
>>> unsigned short z_advise;
>>>
On Fri, Aug 20, 2021 at 05:29:36PM +0800, Chao Yu wrote:
> On 2021/8/20 17:12, Gao Xiang wrote:
> > Hi Chao,
> >
> > On Fri, Aug 20, 2021 at 05:04:13PM +0800, Chao Yu wrote:
> > > On 2021/8/19 14:33, Gao Xiang wrote:
> >
> > ...
> >
> > > > }
> > > > +static int erofs_map_blocks(struct inode *inode,
> > > > + struct erofs_map_blocks *map, int flags)
> > > > +{
> > > > + struct super_block *sb = inode->i_sb;
> > > > + struct erofs_inode *vi = EROFS_I(inode);
> > > > + struct erofs_inode_chunk_index *idx;
> > > > + struct page *page;
> > > > + u64 chunknr;
> > > > + unsigned int unit;
> > > > + erofs_off_t pos;
> > > > + int err = 0;
> > > > +
> > > > + if (map->m_la >= inode->i_size) {
> > > > + /* leave out-of-bound access unmapped */
> > > > + map->m_flags = 0;
> > > > + map->m_plen = 0;
> > > > + goto out;
> > > > + }
> > > > +
> > > > + if (vi->datalayout != EROFS_INODE_CHUNK_BASED)
> > > > + return erofs_map_blocks_flatmode(inode, map, flags);
> > > > +
> > > > + if (vi->chunkformat & EROFS_CHUNK_FORMAT_INDEXES)
> > > > + unit = sizeof(*idx); /* chunk index */
> > > > + else
> > > > + unit = 4; /* block map */
> > >
> > > You mean sizeof(__le32)?
> >
> > Yeah, sizeof(__le32) == 4, either way works for me.
> >
> > If some tendency about this, I will update when applying.
>
> Xiang,
>
> Yeah, I preper:
>
> #define EROFS_BLOCK_MAP_ENTRY_SIZE sizeof(__le32)
>
> unit = EROFS_BLOCK_MAP_ENTRY_SIZE;
>
> to improve readablity, but unit = sizeof(__le32) is fine as well.
Ok, looks much better, let me revise v3 here.
Thanks,
Gao Xiang
>
> Thanks,
>
> >
> > >
> > > Otherwise it looks good to me.
> > >
> > > Reviewed-by: Chao Yu <[email protected]>
> > >
> >
> > Thanks for the review!
> >
> > Thanks,
> > Gao Xiang
> >
> > > Thanks,
> > >
Currently, uncompressed data except for tail-packing inline is
consecutive on disk.
In order to support chunk-based data deduplication, add a new
corresponding inode data layout.
In the future, the data source of chunks can be either (un)compressed.
Reviewed-by: Liu Bo <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Signed-off-by: Gao Xiang <[email protected]>
---
changes since v2:
- introduce EROFS_BLOCK_MAP_ENTRY_SIZE suggested by Chao.
Documentation/filesystems/erofs.rst | 16 ++++++++--
fs/erofs/erofs_fs.h | 47 +++++++++++++++++++++++++++--
2 files changed, 59 insertions(+), 4 deletions(-)
diff --git a/Documentation/filesystems/erofs.rst b/Documentation/filesystems/erofs.rst
index 868e3972227f..d484408a90c1 100644
--- a/Documentation/filesystems/erofs.rst
+++ b/Documentation/filesystems/erofs.rst
@@ -156,13 +156,14 @@ may not. All metadatas can be now observed in two different spaces (views):
Xattrs, extents, data inline are followed by the corresponding inode with
proper alignment, and they could be optional for different data mappings.
- _currently_ total 4 valid data mappings are supported:
+ _currently_ total 5 data layouts are supported:
== ====================================================================
0 flat file data without data inline (no extent);
1 fixed-sized output data compression (with non-compacted indexes);
2 flat file data with tail packing data inline (no extent);
- 3 fixed-sized output data compression (with compacted indexes, v5.3+).
+ 3 fixed-sized output data compression (with compacted indexes, v5.3+);
+ 4 chunk-based file (v5.15+).
== ====================================================================
The size of the optional xattrs is indicated by i_xattr_count in inode
@@ -213,6 +214,17 @@ Note that apart from the offset of the first filename, nameoff0 also indicates
the total number of directory entries in this block since it is no need to
introduce another on-disk field at all.
+Chunk-based file
+----------------
+In order to support chunk-based data deduplication, a new inode data layout has
+been supported since Linux v5.15: Files are split in equal-sized data chunks
+with ``extents`` area of the inode metadata indicating how to get the chunk
+data: these can be simply as a 4-byte block address array or in the 8-byte
+chunk index form (see struct erofs_inode_chunk_index in erofs_fs.h for more
+details.)
+
+By the way, chunk-based files are all uncompressed for now.
+
Data compression
----------------
EROFS implements LZ4 fixed-sized output compression which generates fixed-sized
diff --git a/fs/erofs/erofs_fs.h b/fs/erofs/erofs_fs.h
index 0f8da74570b4..b0b23f41abc3 100644
--- a/fs/erofs/erofs_fs.h
+++ b/fs/erofs/erofs_fs.h
@@ -4,6 +4,7 @@
*
* Copyright (C) 2017-2018 HUAWEI, Inc.
* https://www.huawei.com/
+ * Copyright (C) 2021, Alibaba Cloud
*/
#ifndef __EROFS_FS_H
#define __EROFS_FS_H
@@ -19,10 +20,12 @@
#define EROFS_FEATURE_INCOMPAT_LZ4_0PADDING 0x00000001
#define EROFS_FEATURE_INCOMPAT_COMPR_CFGS 0x00000002
#define EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER 0x00000002
+#define EROFS_FEATURE_INCOMPAT_CHUNKED_FILE 0x00000004
#define EROFS_ALL_FEATURE_INCOMPAT \
(EROFS_FEATURE_INCOMPAT_LZ4_0PADDING | \
EROFS_FEATURE_INCOMPAT_COMPR_CFGS | \
- EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER)
+ EROFS_FEATURE_INCOMPAT_BIG_PCLUSTER | \
+ EROFS_FEATURE_INCOMPAT_CHUNKED_FILE)
#define EROFS_SB_EXTSLOT_SIZE 16
@@ -64,13 +67,16 @@ struct erofs_super_block {
* inode, [xattrs], last_inline_data, ... | ... | no-holed data
* 3 - inode compression D:
* inode, [xattrs], map_header, extents ... | ...
- * 4~7 - reserved
+ * 4 - inode chunk-based E:
+ * inode, [xattrs], chunk indexes ... | ...
+ * 5~7 - reserved
*/
enum {
EROFS_INODE_FLAT_PLAIN = 0,
EROFS_INODE_FLAT_COMPRESSION_LEGACY = 1,
EROFS_INODE_FLAT_INLINE = 2,
EROFS_INODE_FLAT_COMPRESSION = 3,
+ EROFS_INODE_CHUNK_BASED = 4,
EROFS_INODE_DATALAYOUT_MAX
};
@@ -90,6 +96,19 @@ static inline bool erofs_inode_is_data_compressed(unsigned int datamode)
#define EROFS_I_ALL \
((1 << (EROFS_I_DATALAYOUT_BIT + EROFS_I_DATALAYOUT_BITS)) - 1)
+/* indicate chunk blkbits, thus 'chunksize = blocksize << chunk blkbits' */
+#define EROFS_CHUNK_FORMAT_BLKBITS_MASK 0x001F
+/* with chunk indexes or just a 4-byte blkaddr array */
+#define EROFS_CHUNK_FORMAT_INDEXES 0x0020
+
+#define EROFS_CHUNK_FORMAT_ALL \
+ (EROFS_CHUNK_FORMAT_BLKBITS_MASK | EROFS_CHUNK_FORMAT_INDEXES)
+
+struct erofs_inode_chunk_info {
+ __le16 format; /* chunk blkbits, etc. */
+ __le16 reserved;
+};
+
/* 32-byte reduced form of an ondisk inode */
struct erofs_inode_compact {
__le16 i_format; /* inode format hints */
@@ -107,6 +126,9 @@ struct erofs_inode_compact {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
__le32 i_ino; /* only used for 32-bit stat compatibility */
__le16 i_uid;
@@ -135,6 +157,9 @@ struct erofs_inode_extended {
/* for device files, used to indicate old/new device # */
__le32 rdev;
+
+ /* for chunk-based files, it contains the summary info */
+ struct erofs_inode_chunk_info c;
} i_u;
/* only used for 32-bit stat compatibility */
@@ -204,6 +229,19 @@ static inline unsigned int erofs_xattr_entry_size(struct erofs_xattr_entry *e)
e->e_name_len + le16_to_cpu(e->e_value_size));
}
+/* represent a zeroed chunk (hole) */
+#define EROFS_NULL_ADDR -1
+
+/* 4-byte block address array */
+#define EROFS_BLOCK_MAP_ENTRY_SIZE sizeof(__le32)
+
+/* 8-byte inode chunk indexes */
+struct erofs_inode_chunk_index {
+ __le16 advise; /* always 0, don't care for now */
+ __le16 device_id; /* back-end storage id, always 0 for now */
+ __le32 blkaddr; /* start block address of this inode chunk */
+};
+
/* maximum supported size of a physical compression cluster */
#define Z_EROFS_PCLUSTER_MAX_SIZE (1024 * 1024)
@@ -338,9 +376,14 @@ static inline void erofs_check_ondisk_layout_definitions(void)
BUILD_BUG_ON(sizeof(struct erofs_inode_extended) != 64);
BUILD_BUG_ON(sizeof(struct erofs_xattr_ibody_header) != 12);
BUILD_BUG_ON(sizeof(struct erofs_xattr_entry) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_info) != 4);
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_map_header) != 8);
BUILD_BUG_ON(sizeof(struct z_erofs_vle_decompressed_index) != 8);
BUILD_BUG_ON(sizeof(struct erofs_dirent) != 12);
+ /* keep in sync between 2 index structures for better extendibility */
+ BUILD_BUG_ON(sizeof(struct erofs_inode_chunk_index) !=
+ sizeof(struct z_erofs_vle_decompressed_index));
BUILD_BUG_ON(BIT(Z_EROFS_VLE_DI_CLUSTER_TYPE_BITS) <
Z_EROFS_VLE_CLUSTER_TYPE_MAX - 1);
--
2.24.4