linux/fs/ext4/mballoc.h
Jan Kara 83e80a6e35 ext4: use buckets for cr 1 block scan instead of rbtree
Using rbtree for sorting groups by average fragment size is relatively
expensive (needs rbtree update on every block freeing or allocation) and
leads to wide spreading of allocations because selection of block group
is very sentitive both to changes in free space and amount of blocks
allocated. Furthermore selecting group with the best matching average
fragment size is not necessary anyway, even more so because the
variability of fragment sizes within a group is likely large so average
is not telling much. We just need a group with large enough average
fragment size so that we have high probability of finding large enough
free extent and we don't want average fragment size to be too big so
that we are likely to find free extent only somewhat larger than what we
need.

So instead of maintaing rbtree of groups sorted by fragment size keep
bins (lists) or groups where average fragment size is in the interval
[2^i, 2^(i+1)). This structure requires less updates on block allocation
/ freeing, generally avoids chaotic spreading of allocations into block
groups, and still is able to quickly (even faster that the rbtree)
provide a block group which is likely to have a suitably sized free
space extent.

This patch reduces number of block groups used when untarring archive
with medium sized files (size somewhat above 64k which is default
mballoc limit for avoiding locality group preallocation) to about half
and thus improves write speeds for eMMC flash significantly.

Fixes: 196e402adf ("ext4: improve cr 0 / cr 1 group scanning")
CC: stable@kernel.org
Reported-and-tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Tested-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Ritesh Harjani (IBM) <ritesh.list@gmail.com>
Link: https://lore.kernel.org/all/0d81a7c2-46b7-6010-62a4-3e6cfc1628d6@i2se.com/
Link: https://lore.kernel.org/r/20220908092136.11770-5-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2022-09-21 22:12:03 -04:00

238 lines
5.9 KiB
C

// SPDX-License-Identifier: GPL-2.0
/*
* fs/ext4/mballoc.h
*
* Written by: Alex Tomas <alex@clusterfs.com>
*
*/
#ifndef _EXT4_MBALLOC_H
#define _EXT4_MBALLOC_H
#include <linux/time.h>
#include <linux/fs.h>
#include <linux/namei.h>
#include <linux/quotaops.h>
#include <linux/buffer_head.h>
#include <linux/module.h>
#include <linux/swap.h>
#include <linux/proc_fs.h>
#include <linux/pagemap.h>
#include <linux/seq_file.h>
#include <linux/blkdev.h>
#include <linux/mutex.h>
#include "ext4_jbd2.h"
#include "ext4.h"
/*
* mb_debug() dynamic printk msgs could be used to debug mballoc code.
*/
#ifdef CONFIG_EXT4_DEBUG
#define mb_debug(sb, fmt, ...) \
pr_debug("[%s/%d] EXT4-fs (%s): (%s, %d): %s: " fmt, \
current->comm, task_pid_nr(current), sb->s_id, \
__FILE__, __LINE__, __func__, ##__VA_ARGS__)
#else
#define mb_debug(sb, fmt, ...) no_printk(fmt, ##__VA_ARGS__)
#endif
#define EXT4_MB_HISTORY_ALLOC 1 /* allocation */
#define EXT4_MB_HISTORY_PREALLOC 2 /* preallocated blocks used */
/*
* How long mballoc can look for a best extent (in found extents)
*/
#define MB_DEFAULT_MAX_TO_SCAN 200
/*
* How long mballoc must look for a best extent
*/
#define MB_DEFAULT_MIN_TO_SCAN 10
/*
* with 'ext4_mb_stats' allocator will collect stats that will be
* shown at umount. The collecting costs though!
*/
#define MB_DEFAULT_STATS 0
/*
* files smaller than MB_DEFAULT_STREAM_THRESHOLD are served
* by the stream allocator, which purpose is to pack requests
* as close each to other as possible to produce smooth I/O traffic
* We use locality group prealloc space for stream request.
* We can tune the same via /proc/fs/ext4/<partition>/stream_req
*/
#define MB_DEFAULT_STREAM_THRESHOLD 16 /* 64K */
/*
* for which requests use 2^N search using buddies
*/
#define MB_DEFAULT_ORDER2_REQS 2
/*
* default group prealloc size 512 blocks
*/
#define MB_DEFAULT_GROUP_PREALLOC 512
/*
* maximum length of inode prealloc list
*/
#define MB_DEFAULT_MAX_INODE_PREALLOC 512
/*
* Number of groups to search linearly before performing group scanning
* optimization.
*/
#define MB_DEFAULT_LINEAR_LIMIT 4
/*
* Minimum number of groups that should be present in the file system to perform
* group scanning optimizations.
*/
#define MB_DEFAULT_LINEAR_SCAN_THRESHOLD 16
/*
* Number of valid buddy orders
*/
#define MB_NUM_ORDERS(sb) ((sb)->s_blocksize_bits + 2)
struct ext4_free_data {
/* this links the free block information from sb_info */
struct list_head efd_list;
/* this links the free block information from group_info */
struct rb_node efd_node;
/* group which free block extent belongs */
ext4_group_t efd_group;
/* free block extent */
ext4_grpblk_t efd_start_cluster;
ext4_grpblk_t efd_count;
/* transaction which freed this extent */
tid_t efd_tid;
};
struct ext4_prealloc_space {
struct list_head pa_inode_list;
struct list_head pa_group_list;
union {
struct list_head pa_tmp_list;
struct rcu_head pa_rcu;
} u;
spinlock_t pa_lock;
atomic_t pa_count;
unsigned pa_deleted;
ext4_fsblk_t pa_pstart; /* phys. block */
ext4_lblk_t pa_lstart; /* log. block */
ext4_grpblk_t pa_len; /* len of preallocated chunk */
ext4_grpblk_t pa_free; /* how many blocks are free */
unsigned short pa_type; /* pa type. inode or group */
spinlock_t *pa_obj_lock;
struct inode *pa_inode; /* hack, for history only */
};
enum {
MB_INODE_PA = 0,
MB_GROUP_PA = 1
};
struct ext4_free_extent {
ext4_lblk_t fe_logical;
ext4_grpblk_t fe_start; /* In cluster units */
ext4_group_t fe_group;
ext4_grpblk_t fe_len; /* In cluster units */
};
/*
* Locality group:
* we try to group all related changes together
* so that writeback can flush/allocate them together as well
* Size of lg_prealloc_list hash is determined by MB_DEFAULT_GROUP_PREALLOC
* (512). We store prealloc space into the hash based on the pa_free blocks
* order value.ie, fls(pa_free)-1;
*/
#define PREALLOC_TB_SIZE 10
struct ext4_locality_group {
/* for allocator */
/* to serialize allocates */
struct mutex lg_mutex;
/* list of preallocations */
struct list_head lg_prealloc_list[PREALLOC_TB_SIZE];
spinlock_t lg_prealloc_lock;
};
struct ext4_allocation_context {
struct inode *ac_inode;
struct super_block *ac_sb;
/* original request */
struct ext4_free_extent ac_o_ex;
/* goal request (normalized ac_o_ex) */
struct ext4_free_extent ac_g_ex;
/* the best found extent */
struct ext4_free_extent ac_b_ex;
/* copy of the best found extent taken before preallocation efforts */
struct ext4_free_extent ac_f_ex;
__u32 ac_groups_considered;
__u32 ac_flags; /* allocation hints */
__u16 ac_groups_scanned;
__u16 ac_groups_linear_remaining;
__u16 ac_found;
__u16 ac_tail;
__u16 ac_buddy;
__u8 ac_status;
__u8 ac_criteria;
__u8 ac_2order; /* if request is to allocate 2^N blocks and
* N > 0, the field stores N, otherwise 0 */
__u8 ac_op; /* operation, for history only */
struct page *ac_bitmap_page;
struct page *ac_buddy_page;
struct ext4_prealloc_space *ac_pa;
struct ext4_locality_group *ac_lg;
};
#define AC_STATUS_CONTINUE 1
#define AC_STATUS_FOUND 2
#define AC_STATUS_BREAK 3
struct ext4_buddy {
struct page *bd_buddy_page;
void *bd_buddy;
struct page *bd_bitmap_page;
void *bd_bitmap;
struct ext4_group_info *bd_info;
struct super_block *bd_sb;
__u16 bd_blkbits;
ext4_group_t bd_group;
};
static inline ext4_fsblk_t ext4_grp_offs_to_block(struct super_block *sb,
struct ext4_free_extent *fex)
{
return ext4_group_first_block_no(sb, fex->fe_group) +
(fex->fe_start << EXT4_SB(sb)->s_cluster_bits);
}
typedef int (*ext4_mballoc_query_range_fn)(
struct super_block *sb,
ext4_group_t agno,
ext4_grpblk_t start,
ext4_grpblk_t len,
void *priv);
int
ext4_mballoc_query_range(
struct super_block *sb,
ext4_group_t agno,
ext4_grpblk_t start,
ext4_grpblk_t end,
ext4_mballoc_query_range_fn formatter,
void *priv);
#endif