Minki/linux
1
0
Fork 1
mirror of https://github.com/torvalds/linux.git synced 2024-12-04 18:13:04 +00:00
Code Issues Packages Projects Releases Wiki Activity
5f221c3ca1
linux/fs/btrfs/zlib.c

410 lines
10 KiB
C
Raw Normal View History

Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
/*
* Copyright (C) 2008 Oracle. All rights reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License v2 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
* Based on jffs2 zlib code:
* Copyright © 2001-2007 Red Hat, Inc.
* Created by David Woodhouse <dwmw2@infradead.org>
*/
#include <linux/kernel.h>
#include <linux/slab.h>
#include <linux/zlib.h>
#include <linux/zutil.h>
#include <linux/vmalloc.h>
#include <linux/init.h>
#include <linux/err.h>
#include <linux/sched.h>
#include <linux/pagemap.h>
#include <linux/bio.h>
Btrfs: make things static and include the right headers Shut up various sparse warnings about symbols that should be either static or have their declarations in scope. Signed-off-by: Christoph Hellwig <hch@lst.de>
2008-12-02 14:54:17 +00:00
#include "compression.h"
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
struct workspace {
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
z_stream strm;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
char *buf;
struct list_head list;
};
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
static void zlib_free_workspace(struct list_head *ws)
{
struct workspace *workspace = list_entry(ws, struct workspace, list);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
vfree(workspace->strm.workspace);
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
kfree(workspace->buf);
kfree(workspace);
}
static struct list_head *zlib_alloc_workspace(void)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
{
struct workspace *workspace;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
int workspacesize;
btrfs: Fix bugs in zlib workspace - Fix a race that can result in alloc_workspace > cpus. - Fix to check num_workspace after wakeup. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-10-25 07:11:43 +00:00
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
workspace = kzalloc(sizeof(*workspace), GFP_NOFS);
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
if (!workspace)
return ERR_PTR(-ENOMEM);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspacesize = max(zlib_deflate_workspacesize(MAX_WBITS, MAX_MEM_LEVEL),
zlib_inflate_workspacesize());
workspace->strm.workspace = vmalloc(workspacesize);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->buf = kmalloc(PAGE_SIZE, GFP_NOFS);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (!workspace->strm.workspace || !workspace->buf)
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
goto fail;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
INIT_LIST_HEAD(&workspace->list);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
return &workspace->list;
fail:
zlib_free_workspace(&workspace->list);
return ERR_PTR(-ENOMEM);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
static int zlib_compress_pages(struct list_head *ws,
struct address_space *mapping,
btrfs: merge length input and output parameter in compress_pages The length parameter is basically duplicated for input and output in the top level caller of the compress_pages chain. We can simply use one variable for that and reduce stack consumption. The compression implementation will sink the parameter to a local variable so everything works as before. Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-14 18:04:07 +00:00
u64 start,
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
struct page **pages,
unsigned long *out_pages,
unsigned long *total_in,
btrfs: derive maximum output size in the compression implementation The value of max_out can be calculated from the parameters passed to the compressors, which is number of pages and the page size, and we don't have to needlessly pass it around. Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-14 18:45:05 +00:00
unsigned long *total_out)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
{
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
struct workspace *workspace = list_entry(ws, struct workspace, list);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
int ret;
char *data_in;
char *cpage_out;
int nr_pages = 0;
struct page *in_page = NULL;
struct page *out_page = NULL;
unsigned long bytes_left;
btrfs: merge length input and output parameter in compress_pages The length parameter is basically duplicated for input and output in the top level caller of the compress_pages chain. We can simply use one variable for that and reduce stack consumption. The compression implementation will sink the parameter to a local variable so everything works as before. Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-14 18:04:07 +00:00
unsigned long len = *total_out;
btrfs: merge nr_pages input and output parameter in compress_pages The parameter saying how many pages can be allocated at maximum can be merged with the output page counter, to save some stack space. The compression implementation will sink the parameter to a local variable so everything works as before. The nr_pages variables can also be simply merged in compress_file_range into one. Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-14 18:04:07 +00:00
unsigned long nr_dest_pages = *out_pages;
btrfs: derive maximum output size in the compression implementation The value of max_out can be calculated from the parameters passed to the compressors, which is number of pages and the page size, and we don't have to needlessly pass it around. Signed-off-by: David Sterba <dsterba@suse.com>
2017-02-14 18:45:05 +00:00
const unsigned long max_out = nr_dest_pages * PAGE_SIZE;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
*out_pages = 0;
*total_out = 0;
*total_in = 0;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (Z_OK != zlib_deflateInit(&workspace->strm, 3)) {
btrfs: convert printk(KERN_* to use pr_* calls This patch converts printk(KERN_* style messages to use the pr_* versions. One side effect is that anything that was KERN_DEBUG is now automatically a dynamic debug message. Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-09-20 14:05:01 +00:00
pr_warn("BTRFS: deflateInit failed\n");
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.total_in = 0;
workspace->strm.total_out = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
in_page = find_get_page(mapping, start >> PAGE_SHIFT);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
data_in = kmap(in_page);
out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
btrfs: Fix error handling in zlib Return failure if alloc_page() fails to allocate memory, and the upper code will just give up compression. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-09 00:27:27 +00:00
if (out_page == NULL) {
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -ENOMEM;
btrfs: Fix error handling in zlib Return failure if alloc_page() fails to allocate memory, and the upper code will just give up compression. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-09 00:27:27 +00:00
goto out;
}
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
cpage_out = kmap(out_page);
pages[0] = out_page;
nr_pages = 1;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in = data_in;
workspace->strm.next_out = cpage_out;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
workspace->strm.avail_in = min(len, PAGE_SIZE);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
while (workspace->strm.total_in < len) {
ret = zlib_deflate(&workspace->strm, Z_SYNC_FLUSH);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
if (ret != Z_OK) {
btrfs: convert printk(KERN_* to use pr_* calls This patch converts printk(KERN_* style messages to use the pr_* versions. One side effect is that anything that was KERN_DEBUG is now automatically a dynamic debug message. Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-09-20 14:05:01 +00:00
pr_debug("BTRFS: deflate in loop returned %d\n",
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
ret);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
zlib_deflateEnd(&workspace->strm);
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
/* we're making it bigger, give up */
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.total_in > 8192 &&
workspace->strm.total_in <
workspace->strm.total_out) {
btrfs: use E2BIG instead of EIO if compression does not help Return codes got updated in 60e1975acb48fc3d74a3422b21dde74c977ac3d5 (btrfs: return errno instead of -1 from compression) lzo wrapper returns E2BIG in this case, do the same for zlib. Signed-off-by: David Sterba <dsterba@suse.cz>
2014-06-20 09:43:20 +00:00
ret = -E2BIG;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
/* we need another page for writing out. Test this
* before the total_in so we will pull in a new page for
* the stream end if required
*/
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.avail_out == 0) {
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
kunmap(out_page);
if (nr_pages == nr_dest_pages) {
out_page = NULL;
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -E2BIG;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
out_page = alloc_page(GFP_NOFS | __GFP_HIGHMEM);
btrfs: Fix error handling in zlib Return failure if alloc_page() fails to allocate memory, and the upper code will just give up compression. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-09 00:27:27 +00:00
if (out_page == NULL) {
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -ENOMEM;
btrfs: Fix error handling in zlib Return failure if alloc_page() fails to allocate memory, and the upper code will just give up compression. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-09 00:27:27 +00:00
goto out;
}
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
cpage_out = kmap(out_page);
pages[nr_pages] = out_page;
nr_pages++;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_out = cpage_out;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
/* we're all done */
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.total_in >= len)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
/* we've read in a full page, get a new one */
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.avail_in == 0) {
if (workspace->strm.total_out > max_out)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
bytes_left = len - workspace->strm.total_in;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
kunmap(in_page);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(in_page);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
start += PAGE_SIZE;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
in_page = find_get_page(mapping,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
start >> PAGE_SHIFT);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
data_in = kmap(in_page);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.avail_in = min(bytes_left,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
PAGE_SIZE);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in = data_in;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.avail_in = 0;
ret = zlib_deflate(&workspace->strm, Z_FINISH);
zlib_deflateEnd(&workspace->strm);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
if (ret != Z_STREAM_END) {
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.total_out >= workspace->strm.total_in) {
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -E2BIG;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto out;
}
ret = 0;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
*total_out = workspace->strm.total_out;
*total_in = workspace->strm.total_in;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
out:
*out_pages = nr_pages;
if (out_page)
kunmap(out_page);
if (in_page) {
kunmap(in_page);
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
put_page(in_page);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
return ret;
}
btrfs: use bio iterators for the decompression handlers Pass the full bio to the decompression routines and use bio iterators to iterate over the data in the bio. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-25 08:07:46 +00:00
static int zlib_decompress_bio(struct list_head *ws, struct page **pages_in,
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
u64 disk_start,
btrfs: use bio iterators for the decompression handlers Pass the full bio to the decompression routines and use bio iterators to iterate over the data in the bio. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-25 08:07:46 +00:00
struct bio *orig_bio,
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
size_t srclen)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
{
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
struct workspace *workspace = list_entry(ws, struct workspace, list);
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
int ret = 0, ret2;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
int wbits = MAX_WBITS;
char *data_in;
size_t total_out = 0;
unsigned long page_in_index = 0;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
unsigned long total_pages_in = DIV_ROUND_UP(srclen, PAGE_SIZE);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
unsigned long buf_start;
data_in = kmap(pages_in[page_in_index]);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in = data_in;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_in = min_t(size_t, srclen, PAGE_SIZE);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.total_in = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.total_out = 0;
workspace->strm.next_out = workspace->buf;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
/* If it's deflate, and it's got no preset dictionary, then
we can tell zlib to skip the adler32 check. */
if (srclen > 2 && !(data_in[1] & PRESET_DICT) &&
((data_in[0] & 0x0f) == Z_DEFLATED) &&
!(((data_in[0]<<8) + data_in[1]) % 31)) {
wbits = -((data_in[0] >> 4) + 8);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in += 2;
workspace->strm.avail_in -= 2;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (Z_OK != zlib_inflateInit2(&workspace->strm, wbits)) {
btrfs: convert printk(KERN_* to use pr_* calls This patch converts printk(KERN_* style messages to use the pr_* versions. One side effect is that anything that was KERN_DEBUG is now automatically a dynamic debug message. Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-09-20 14:05:01 +00:00
pr_warn("BTRFS: inflateInit failed\n");
btrfs: Call kunmap if zlib_inflateInit2 fails If zlib_inflateInit2 fails, the input page is never unmapped. Add a call to kunmap when it fails. Signed-off-by: Nick Terrell <nickrterrell@gmail.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-02 03:25:27 +00:00
kunmap(pages_in[page_in_index]);
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
return -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
while (workspace->strm.total_in < srclen) {
ret = zlib_inflate(&workspace->strm, Z_NO_FLUSH);
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (ret != Z_OK && ret != Z_STREAM_END)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
buf_start = total_out;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
total_out = workspace->strm.total_out;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
/* we didn't make progress in this inflate call, we're done */
if (buf_start == total_out)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
ret2 = btrfs_decompress_buf2page(workspace->buf, buf_start,
total_out, disk_start,
btrfs: use bio iterators for the decompression handlers Pass the full bio to the decompression routines and use bio iterators to iterate over the data in the bio. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-25 08:07:46 +00:00
orig_bio);
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
if (ret2 == 0) {
ret = 0;
goto done;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
btrfs: Extract duplicate decompress code Add a common function to copy decompressed data from working buffer to bio pages. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-11-08 07:22:19 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_out = workspace->buf;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (workspace->strm.avail_in == 0) {
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
unsigned long tmp;
kunmap(pages_in[page_in_index]);
page_in_index++;
if (page_in_index >= total_pages_in) {
data_in = NULL;
break;
}
data_in = kmap(pages_in[page_in_index]);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in = data_in;
tmp = srclen - workspace->strm.total_in;
workspace->strm.avail_in = min(tmp,
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
PAGE_SIZE);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
}
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (ret != Z_STREAM_END)
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
else
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
ret = 0;
done:
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
zlib_inflateEnd(&workspace->strm);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
if (data_in)
kunmap(pages_in[page_in_index]);
btrfs: zero out left over bytes after processing compression streams Don Bailey noticed that our page zeroing for compression at end-io time isn't complete. This reworks a patch from Linus to push the zeroing into the zlib and lzo specific functions instead of trying to handle the corners inside btrfs_decompress_buf2page Signed-off-by: Chris Mason <clm@fb.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Reported-by: Don A. Bailey <donb@securitymouse.com> cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-30 13:56:33 +00:00
if (!ret)
btrfs: use bio iterators for the decompression handlers Pass the full bio to the decompression routines and use bio iterators to iterate over the data in the bio. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-25 08:07:46 +00:00
zero_fill_bio(orig_bio);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
return ret;
}
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
static int zlib_decompress(struct list_head *ws, unsigned char *data_in,
struct page *dest_page,
unsigned long start_byte,
size_t srclen, size_t destlen)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
{
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
struct workspace *workspace = list_entry(ws, struct workspace, list);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
int ret = 0;
int wbits = MAX_WBITS;
btrfs: zero out left over bytes after processing compression streams Don Bailey noticed that our page zeroing for compression at end-io time isn't complete. This reworks a patch from Linus to push the zeroing into the zlib and lzo specific functions instead of trying to handle the corners inside btrfs_decompress_buf2page Signed-off-by: Chris Mason <clm@fb.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Reported-by: Don A. Bailey <donb@securitymouse.com> cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-30 13:56:33 +00:00
unsigned long bytes_left;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
unsigned long total_out = 0;
btrfs: zero out left over bytes after processing compression streams Don Bailey noticed that our page zeroing for compression at end-io time isn't complete. This reworks a patch from Linus to push the zeroing into the zlib and lzo specific functions instead of trying to handle the corners inside btrfs_decompress_buf2page Signed-off-by: Chris Mason <clm@fb.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Reported-by: Don A. Bailey <donb@securitymouse.com> cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-30 13:56:33 +00:00
unsigned long pg_offset = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
char *kaddr;
btrfs: zero out left over bytes after processing compression streams Don Bailey noticed that our page zeroing for compression at end-io time isn't complete. This reworks a patch from Linus to push the zeroing into the zlib and lzo specific functions instead of trying to handle the corners inside btrfs_decompress_buf2page Signed-off-by: Chris Mason <clm@fb.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Reported-by: Don A. Bailey <donb@securitymouse.com> cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-30 13:56:33 +00:00
destlen = min_t(unsigned long, destlen, PAGE_SIZE);
bytes_left = destlen;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in = data_in;
workspace->strm.avail_in = srclen;
workspace->strm.total_in = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_out = workspace->buf;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.total_out = 0;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
/* If it's deflate, and it's got no preset dictionary, then
we can tell zlib to skip the adler32 check. */
if (srclen > 2 && !(data_in[1] & PRESET_DICT) &&
((data_in[0] & 0x0f) == Z_DEFLATED) &&
!(((data_in[0]<<8) + data_in[1]) % 31)) {
wbits = -((data_in[0] >> 4) + 8);
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_in += 2;
workspace->strm.avail_in -= 2;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
if (Z_OK != zlib_inflateInit2(&workspace->strm, wbits)) {
btrfs: convert printk(KERN_* to use pr_* calls This patch converts printk(KERN_* style messages to use the pr_* versions. One side effect is that anything that was KERN_DEBUG is now automatically a dynamic debug message. Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-09-20 14:05:01 +00:00
pr_warn("BTRFS: inflateInit failed\n");
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
return -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
while (bytes_left > 0) {
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
unsigned long buf_start;
unsigned long buf_offset;
unsigned long bytes;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
ret = zlib_inflate(&workspace->strm, Z_NO_FLUSH);
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (ret != Z_OK && ret != Z_STREAM_END)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
buf_start = total_out;
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
total_out = workspace->strm.total_out;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
if (total_out == buf_start) {
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
break;
}
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (total_out <= start_byte)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
goto next;
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (total_out > start_byte && buf_start < start_byte)
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
buf_offset = start_byte - buf_start;
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
else
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
buf_offset = 0;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
bytes = min(PAGE_SIZE - pg_offset,
PAGE_SIZE - buf_offset);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
bytes = min(bytes, bytes_left);
btrfs: remove the second argument of k[un]map_atomic() Signed-off-by: Cong Wang <amwang@redhat.com>
2011-11-25 15:14:28 +00:00
kaddr = kmap_atomic(dest_page);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
memcpy(kaddr + pg_offset, workspace->buf + buf_offset, bytes);
btrfs: remove the second argument of k[un]map_atomic() Signed-off-by: Cong Wang <amwang@redhat.com>
2011-11-25 15:14:28 +00:00
kunmap_atomic(kaddr);
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
pg_offset += bytes;
bytes_left -= bytes;
next:
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
workspace->strm.next_out = workspace->buf;
mm, fs: get rid of PAGE_CACHE_* and page_cache_{get,release} macros PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} macros were introduced *long* time ago with promise that one day it will be possible to implement page cache with bigger chunks than PAGE_SIZE. This promise never materialized. And unlikely will. We have many places where PAGE_CACHE_SIZE assumed to be equal to PAGE_SIZE. And it's constant source of confusion on whether PAGE_CACHE_* or PAGE_* constant should be used in a particular case, especially on the border between fs and mm. Global switching to PAGE_CACHE_SIZE != PAGE_SIZE would cause to much breakage to be doable. Let's stop pretending that pages in page cache are special. They are not. The changes are pretty straight-forward: - <foo> << (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - <foo> >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) -> <foo>; - PAGE_CACHE_{SIZE,SHIFT,MASK,ALIGN} -> PAGE_{SIZE,SHIFT,MASK,ALIGN}; - page_cache_get() -> get_page(); - page_cache_release() -> put_page(); This patch contains automated changes generated with coccinelle using script below. For some reason, coccinelle doesn't patch header files. I've called spatch for them manually. The only adjustment after coccinelle is revert of changes to PAGE_CAHCE_ALIGN definition: we are going to drop it later. There are few places in the code where coccinelle didn't reach. I'll fix them manually in a separate patch. Comments and documentation also will be addressed with the separate patch. virtual patch @@ expression E; @@ - E << (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ expression E; @@ - E >> (PAGE_CACHE_SHIFT - PAGE_SHIFT) + E @@ @@ - PAGE_CACHE_SHIFT + PAGE_SHIFT @@ @@ - PAGE_CACHE_SIZE + PAGE_SIZE @@ @@ - PAGE_CACHE_MASK + PAGE_MASK @@ expression E; @@ - PAGE_CACHE_ALIGN(E) + PAGE_ALIGN(E) @@ expression E; @@ - page_cache_get(E) + get_page(E) @@ expression E; @@ - page_cache_release(E) + put_page(E) Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Acked-by: Michal Hocko <mhocko@suse.com> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-04-01 12:29:47 +00:00
workspace->strm.avail_out = PAGE_SIZE;
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
}
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
if (ret != Z_STREAM_END && bytes_left != 0)
btrfs: return errno instead of -1 from compression The compression layer seems to have been built to return -1 and have callers make up errors that make sense. This isn't great because there are different errors that originate down in the compression layer. Let's return real negative errnos from the compression layer so that callers can pass on the error without having to guess what happened. ENOMEM for allocation failure, E2BIG when compression exceeds the uncompressed input, and EIO for everything else. This helps a future path return errors from btrfs_decompress(). Signed-off-by: Zach Brown <zab@redhat.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-05-09 21:15:08 +00:00
ret = -EIO;
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
else
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
ret = 0;
Btrfs: Fix checkpatch.pl warnings There were many, most are fixed now. struct-funcs.c generates some warnings but these are bogus. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2009-01-06 02:25:51 +00:00
btrfs compression: merge inflate and deflate z_streams `struct workspace' used for zlib compression contains two zlib z_stream-s: `def_strm' used in zlib_compress_pages(), and `inf_strm' used in zlib_decompress/zlib_decompress_biovec(). None of these functions use `inf_strm' and `def_strm' simultaniously, meaning that for every compress/decompress operation we need only one z_stream (out of two available). `inf_strm' and `def_strm' are different in size of ->workspace. For inflate stream we vmalloc() zlib_inflate_workspacesize() bytes, for deflate stream - zlib_deflate_workspacesize() bytes. On my system zlib returns the following workspace sizes, correspondingly: 42312 and 268104 (+ guard pages). Keep only one `z_stream' in `struct workspace' and use it for both compression and decompression. Hence, instead of vmalloc() of two z_stream->worskpace-s, allocate only one of size: max(zlib_deflate_workspacesize(), zlib_inflate_workspacesize()) Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-07 14:38:29 +00:00
zlib_inflateEnd(&workspace->strm);
btrfs: zero out left over bytes after processing compression streams Don Bailey noticed that our page zeroing for compression at end-io time isn't complete. This reworks a patch from Linus to push the zeroing into the zlib and lzo specific functions instead of trying to handle the corners inside btrfs_decompress_buf2page Signed-off-by: Chris Mason <clm@fb.com> Reviewed-by: Josef Bacik <jbacik@fb.com> Reported-by: Don A. Bailey <donb@securitymouse.com> cc: stable@vger.kernel.org Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2014-11-30 13:56:33 +00:00
/*
* this should only happen if zlib returned fewer bytes than we
* expected. btrfs_get_block is responsible for zeroing from the
* end of the inline extent (destlen) to the end of the page
*/
if (pg_offset < destlen) {
kaddr = kmap_atomic(dest_page);
memset(kaddr + pg_offset, 0, destlen - pg_offset);
kunmap_atomic(kaddr);
}
Btrfs: Add zlib compression support This is a large change for adding compression on reading and writing, both for inline and regular extents. It does some fairly large surgery to the writeback paths. Compression is off by default and enabled by mount -o compress. Even when the -o compress mount option is not used, it is possible to read compressed extents off the disk. If compression for a given set of pages fails to make them smaller, the file is flagged to avoid future compression attempts later. * While finding delalloc extents, the pages are locked before being sent down to the delalloc handler. This allows the delalloc handler to do complex things such as cleaning the pages, marking them writeback and starting IO on their behalf. * Inline extents are inserted at delalloc time now. This allows us to compress the data before inserting the inline extent, and it allows us to insert an inline extent that spans multiple pages. * All of the in-memory extent representations (extent_map.c, ordered-data.c etc) are changed to record both an in-memory size and an on disk size, as well as a flag for compression. From a disk format point of view, the extent pointers in the file are changed to record the on disk size of a given extent and some encoding flags. Space in the disk format is allocated for compression encoding, as well as encryption and a generic 'other' field. Neither the encryption or the 'other' field are currently used. In order to limit the amount of data read for a single random read in the file, the size of a compressed extent is limited to 128k. This is a software only limit, the disk format supports u64 sized compressed extents. In order to limit the ram consumed while processing extents, the uncompressed size of a compressed extent is limited to 256k. This is a software only limit and will be subject to tuning later. Checksumming is still done on compressed extents, and it is done on the uncompressed version of the data. This way additional encodings can be layered on without having to figure out which encoding to checksum. Compression happens at delalloc time, which is basically singled threaded because it is usually done by a single pdflush thread. This makes it tricky to spread the compression load across all the cpus on the box. We'll have to look at parallel pdflush walks of dirty inodes at a later time. Decompression is hooked into readpages and it does spread across CPUs nicely. Signed-off-by: Chris Mason <chris.mason@oracle.com>
2008-10-29 18:49:59 +00:00
return ret;
}
btrfs: constify structs with op functions or static definitions There are some op tables that can be easily made const, similarly the sysfs feature and raid tables. This is motivated by PaX CONSTIFY plugin. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-02 17:23:10 +00:00
const struct btrfs_compress_op btrfs_zlib_compress = {
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
.alloc_workspace = zlib_alloc_workspace,
.free_workspace = zlib_free_workspace,
.compress_pages = zlib_compress_pages,
btrfs: use bio iterators for the decompression handlers Pass the full bio to the decompression routines and use bio iterators to iterate over the data in the bio. Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: David Sterba <dsterba@suse.com>
2016-11-25 08:07:46 +00:00
.decompress_bio = zlib_decompress_bio,
btrfs: Allow to add new compression algorithm Make the code aware of compression type, instead of always assuming zlib compression. Also make the zlib workspace function as common code for all compression types. Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
2010-12-17 06:21:50 +00:00
.decompress = zlib_decompress,
};
Reference in New Issue Copy Permalink