experimental branch rebased
Hello everyone,
I've rebased the experimental branch again with my latest performance fixes.
I took out the delayed unlink code, it wasn't making a big enough difference in any benchmarks to justify the complexity.
I changed the delayed backref code to do delayed processing for all extents. In general it is much faster and uses less stack.
I pulled Josef's enospc work into the master branch and asked Linus to pull it.
I'm going to hammer on the experimental branch for a few days and ask Steve to give it another run.
-chris
I have also cleaned up the backport experimental patch so it applies cleanly on the latest code.
Lee
async-thread.c +3 -0
@@ -20,7 +20,10 @@#include <linux/list.h> #include <linux/spinlock.h> #include <linux/freezer.h> +#include <linux/version.h> +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )#include <linux/ftrace.h> +#endif #include "async-thread.h" #define WORK_QUEUED_BIT 0
compat.h +26 -0
@@ -1,7 +1,33 @@#ifndef _COMPAT_H_#define _COMPAT_H_ +#include <linux/version.h> +#define btrfs_drop_nlink (inode) drop_nlink(inode)#define btrfs_inc_nlink (inode) inc_nlink(inode) +#if LINUX_VERSION_CODE <=KERNEL_VERSION (2 ,6 ,27 ) +static inline struct dentry *d_obtain_alias (struct inode *inode) +{ +struct dentry *d; + +if (!inode) +return NULL ; +if (IS_ERR (inode)) +return ERR_CAST (inode); + + d =d_alloc_anon (inode); +if (!d) +iput (inode); +return d; +} +#endif + +#if LINUX_VERSION_CODE <KERNEL_VERSION (2 ,6 ,28 ) +#define __pagevec_lru_add_file __pagevec_lru_add +#define open_bdev_exclusive open_bdev_excl +#define close_bdev_exclusive (bdev, mode) close_bdev_excl(bdev) +typedef unsigned __bitwise__ fmode_t ; +#endif +#endif /* _COMPAT_H_ */
export.h +28 -0
@@ -16,4 +16,32 @@u64 parent_root_objectid; } __attribute__ ((packed)); +// BTRFS needs the btrfs from fid_type which was added in 2.6.27 +// All I did was copy the btrfs defse which was found in +// linux-2.6.27/include/linux/exportfs.h +#if LINUX_VERSION_CODE <KERNEL_VERSION (2 ,6 ,27 ) +enum btrfs_fid_type { +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number. +*/ +FILEID_BTRFS_WITHOUT_PARENT =0x4d , + +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number, +* 64 bit parent object ID, 32 bit parent generation. +*/ +FILEID_BTRFS_WITH_PARENT =0x4e , + +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number, +* 64 bit parent object ID, 32 bit parent generation, +* 64 bit parent root object ID. +*/ +FILEID_BTRFS_WITH_PARENT_ROOT =0x4f , +}; +#endif +#endif
extent_io.c +14 -0
@@ -2847,6 +2847,7 @@return sector; } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )int extent_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len,get_extent_t *get_extent) {
@@ -2938,6 +2939,7 @@ GFP_NOFS); return ret; } +#endif static inline struct page *extent_buffer_page (struct extent_buffer *eb,unsigned long i)
@@ -3163,13 +3165,21 @@ } }clear_page_dirty_for_io (page); +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )spin_lock_irq (&page->mapping->tree_lock); +#else +read_lock_irq (&page->mapping->tree_lock); +#endif if (!PageDirty (page)) {radix_tree_tag_clear (&page->mapping->page_tree,page_index (page),PAGECACHE_TAG_DIRTY ); } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )spin_unlock_irq (&page->mapping->tree_lock); +#else +read_unlock_irq (&page->mapping->tree_lock); +#endif unlock_page (page); }return 0 ;
@@ -3347,7 +3357,11 @@for (i = start_i; i < num_pages; i++) { page =extent_buffer_page (eb, i);if (!wait) { +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )if (!trylock_page (page)) +#else + if(!TestSetPageLocked(page)) +#endif goto unlock_exit; }else {lock_page (page);
extent_io.h +3 -0
@@ -2,6 +2,7 @@#define __EXTENTIO__#include <linux/rbtree.h> +#include <linux/version.h> /* bits for the extent state */ #define EXTENT_DIRTY 1
@@ -205,8 +206,10 @@ unsigned from, unsigned to);sector_t extent_bmap (struct address_space *mapping,sector_t iblock,get_extent_t *get_extent); +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )int extent_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len,get_extent_t *get_extent); +#endif int set_range_dirty (struct extent_io_tree *tree,u64 start,u64 end);int set_state_private (struct extent_io_tree *tree,u64 start,u64 private);int get_state_private (struct extent_io_tree *tree,u64 start,u64 *private);
extent-tree.c +4 -0
@@ -870,7 +870,11 @@static void btrfs_issue_discard (struct block_device *bdev,u64 start,u64 len) { +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,28 )blkdev_issue_discard (bdev, start >>9 , len >>9 ,GFP_KERNEL ); +#else +blkdev_issue_discard (bdev, start >>9 , len >>9 ); +#endif }#endif
file.c +4 -0
@@ -1042,7 +1042,11 @@if (count ==0 ) goto out_nolock; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) err =file_remove_suid (file); +#else + err =remove_suid (fdentry (file)); +#endif if (err) goto out_nolock;file_update_time (file);
inode.c +27 -1
@@ -3456,14 +3456,22 @@if (objectid > root->highest_inode) root->highest_inode = objectid; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) inode->i_uid =current_fsuid (); +#else + inode->i_uid = current->fsuid; +#endif if (dir && (dir->i_mode &S_ISGID )) { inode->i_gid = dir->i_gid;if (S_ISDIR (mode)) mode |=S_ISGID ; } else +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) inode->i_gid =current_fsgid (); +#else + inode->i_gid = current->fsgid; +#endif inode->i_mode = mode; inode->i_ino = objectid;
@@ -4164,11 +4172,13 @@return -EINVAL ; } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )static int btrfs_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len) {return extent_fiemap (inode, fieinfo, start, len, btrfs_get_extent); } +#endif int btrfs_readpage (struct file *file,struct page *page) {
@@ -4500,7 +4510,11 @@kmem_cache_free (btrfs_inode_cachep,BTRFS_I (inode)); } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) static void init_once(void *foo) +#else +static void init_once(struct kmem_cache *bar, void *foo) +#endif {struct btrfs_inode *ei = (struct btrfs_inode *) foo;
@@ -4523,7 +4537,12 @@ struct kmem_cache *btrfs_cache_create(const char *name, size_t size, unsigned long extra_flags, -void (*ctor)(void *)) +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27) + void (*ctor)(void *) +#else +void (*ctor)(struct kmem_cache *,void *) +#endif +) {return kmem_cache_create (name, size,0 , (SLAB_RECLAIM_ACCOUNT |SLAB_MEM_SPREAD | extra_flags), ctor);
@@ -4942,7 +4961,12 @@return __set_page_dirty_nobuffers (page); } +#if LINUX_VERSION_CODE >KERNEL_VERSION (2 ,6 ,26 ) static int btrfs_permission(struct inode *inode, int mask) +#else +static int btrfs_permission (struct inode *inode,int mask, + struct nameidata *nd) +#endif {if (btrfs_test_flag (inode,READONLY ) && (mask &MAY_WRITE ))return -EACCES ;
@@ -5036,7 +5060,9 @@ .removexattr = btrfs_removexattr, .permission = btrfs_permission, .fallocate = btrfs_fallocate, +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 ) .fiemap = btrfs_fiemap, +#endif };static struct inode_operations btrfs_special_inode_operations = { .getattr = btrfs_getattr,
ioctl.c +4 -0
@@ -240,7 +240,11 @@return -EEXIST ;if (IS_DEADDIR (dir))return -ENOENT ; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )return inode_permission (dir,MAY_WRITE |MAY_EXEC ); +#else +return permission (dir,MAY_WRITE |MAY_EXEC ,NULL ); +#endif }/*
super.c +33 -0
@@ -38,6 +38,7 @@#include <linux/namei.h> #include <linux/miscdevice.h> #include <linux/magic.h> +#include <linux/version.h> #include "compat.h" #include "ctree.h" #include "disk-io.h"
@@ -51,6 +52,9 @@#include "export.h" #include "compression.h" +#ifndef BTRFS_SUPER_MAGIC +#define BTRFS_SUPER_MAGIC 0x9123683E +#endif static struct super_operations btrfs_super_ops;
@@ -609,6 +613,8 @@return ret; } + +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )static int btrfs_freeze (struct super_block *sb) {struct btrfs_root *root =btrfs_sb (sb);
@@ -617,6 +623,7 @@return 0 ; } +static int btrfs_unfreeze (struct super_block *sb) {struct btrfs_root *root =btrfs_sb (sb);
@@ -625,6 +632,27 @@return 0 ; } + +#else + + +static void btrfs_write_super_lockfs (struct super_block *sb) +{ +struct btrfs_root *root =btrfs_sb (sb); +mutex_lock (&root->fs_info->transaction_kthread_mutex); +mutex_lock (&root->fs_info->cleaner_mutex); +} + + +static void btrfs_unlockfs (struct super_block *sb) +{ +struct btrfs_root *root =btrfs_sb (sb); +mutex_unlock (&root->fs_info->cleaner_mutex); +mutex_unlock (&root->fs_info->transaction_kthread_mutex); +} +#endif + +static struct super_operations btrfs_super_ops = { .delete_inode = btrfs_delete_inode, .put_super = btrfs_put_super,
@@ -637,8 +665,13 @@ .destroy_inode = btrfs_destroy_inode, .statfs = btrfs_statfs, .remount_fs = btrfs_remount, +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 ) .freeze_fs = btrfs_freeze, .unfreeze_fs = btrfs_unfreeze, +#else + .write_super_lockfs = btrfs_write_super_lockfs, + .unlockfs = btrfs_unlockfs, +#endif };static const struct file_operations btrfs_ctl_fops = {
Lee,
The 02/26/2009 version is missing some of the patches that were in the 02/20/2009 version so it will not compile on 26.
I found the bug in the patch that breaks on 26:
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27)
if (!trylock_page(page))
+#else
+ if(!TestSetPageLocked(page))
+#endif
It should be
if (TestSetPageLocked(page))
because pass/fail is reversed on those functions.
While I was testing, I ported back to 24 on ubuntu and made it work with 2.6.24-23-generic with the experimental tree from feb 25.
I will give you the 24 patches to include in your patch as soon as you have a version that I can rebase against the current experimental on 24/26.
jim
Sorry it took so long to reply its been a crazy week for me. Anyway thanks for figuring this out for me. I can't believe I made such a stupid mistake. Anyway here is the rebased patch.
Lee
async-thread.c +3 -0
@@ -20,7 +20,10 @@#include <linux/list.h> #include <linux/spinlock.h> #include <linux/freezer.h> +#include <linux/version.h> +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )#include <linux/ftrace.h> +#endif #include "async-thread.h" #define WORK_QUEUED_BIT 0
compat.h +26 -0
@@ -1,7 +1,33 @@#ifndef _COMPAT_H_#define _COMPAT_H_ +#include <linux/version.h> +#define btrfs_drop_nlink (inode) drop_nlink(inode)#define btrfs_inc_nlink (inode) inc_nlink(inode) +#if LINUX_VERSION_CODE <=KERNEL_VERSION (2 ,6 ,27 ) +static inline struct dentry *d_obtain_alias (struct inode *inode) +{ +struct dentry *d; + +if (!inode) +return NULL ; +if (IS_ERR (inode)) +return ERR_CAST (inode); + + d =d_alloc_anon (inode); +if (!d) +iput (inode); +return d; +} +#endif + +#if LINUX_VERSION_CODE <KERNEL_VERSION (2 ,6 ,28 ) +#define __pagevec_lru_add_file __pagevec_lru_add +#define open_bdev_exclusive open_bdev_excl +#define close_bdev_exclusive (bdev, mode) close_bdev_excl(bdev) +typedef unsigned __bitwise__ fmode_t ; +#endif +#endif /* _COMPAT_H_ */
export.h +28 -0
@@ -16,4 +16,32 @@u64 parent_root_objectid; } __attribute__ ((packed)); +// BTRFS needs the btrfs from fid_type which was added in 2.6.27 +// All I did was copy the btrfs defse which was found in +// linux-2.6.27/include/linux/exportfs.h +#if LINUX_VERSION_CODE <KERNEL_VERSION (2 ,6 ,27 ) +enum btrfs_fid_type { +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number. +*/ +FILEID_BTRFS_WITHOUT_PARENT =0x4d , + +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number, +* 64 bit parent object ID, 32 bit parent generation. +*/ +FILEID_BTRFS_WITH_PARENT =0x4e , + +/* +* 64 bit object ID, 64 bit root object ID, +* 32 bit generation number, +* 64 bit parent object ID, 32 bit parent generation, +* 64 bit parent root object ID. +*/ +FILEID_BTRFS_WITH_PARENT_ROOT =0x4f , +}; +#endif +#endif
extent_io.c +14 -0
@@ -2847,6 +2847,7 @@return sector; } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )int extent_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len,get_extent_t *get_extent) {
@@ -2938,6 +2939,7 @@ GFP_NOFS); return ret; } +#endif static inline struct page *extent_buffer_page (struct extent_buffer *eb,unsigned long i)
@@ -3163,13 +3165,21 @@ } }clear_page_dirty_for_io (page); +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )spin_lock_irq (&page->mapping->tree_lock); +#else +read_lock_irq (&page->mapping->tree_lock); +#endif if (!PageDirty (page)) {radix_tree_tag_clear (&page->mapping->page_tree,page_index (page),PAGECACHE_TAG_DIRTY ); } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )spin_unlock_irq (&page->mapping->tree_lock); +#else +read_unlock_irq (&page->mapping->tree_lock); +#endif unlock_page (page); }return 0 ;
@@ -3347,7 +3357,11 @@for (i = start_i; i < num_pages; i++) { page =extent_buffer_page (eb, i);if (!wait) { +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )if (!trylock_page (page)) +#else + if(TestSetPageLocked(page)) +#endif goto unlock_exit; }else {lock_page (page);
extent_io.h +3 -0
@@ -2,6 +2,7 @@#define __EXTENTIO__#include <linux/rbtree.h> +#include <linux/version.h> /* bits for the extent state */ #define EXTENT_DIRTY 1
@@ -205,8 +206,10 @@ unsigned from, unsigned to);sector_t extent_bmap (struct address_space *mapping,sector_t iblock,get_extent_t *get_extent); +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )int extent_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len,get_extent_t *get_extent); +#endif int set_range_dirty (struct extent_io_tree *tree,u64 start,u64 end);int set_state_private (struct extent_io_tree *tree,u64 start,u64 private);int get_state_private (struct extent_io_tree *tree,u64 start,u64 *private);
extent-tree.c +4 -0
@@ -870,7 +870,11 @@static void btrfs_issue_discard (struct block_device *bdev,u64 start,u64 len) { +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,28 )blkdev_issue_discard (bdev, start >>9 , len >>9 ,GFP_KERNEL ); +#else +blkdev_issue_discard (bdev, start >>9 , len >>9 ); +#endif }#endif
file.c +4 -0
@@ -1042,7 +1042,11 @@if (count ==0 ) goto out_nolock; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) err =file_remove_suid (file); +#else + err =remove_suid (fdentry (file)); +#endif if (err) goto out_nolock;file_update_time (file);
inode.c +27 -1
@@ -3456,14 +3456,22 @@if (objectid > root->highest_inode) root->highest_inode = objectid; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) inode->i_uid =current_fsuid (); +#else + inode->i_uid = current->fsuid; +#endif if (dir && (dir->i_mode &S_ISGID )) { inode->i_gid = dir->i_gid;if (S_ISDIR (mode)) mode |=S_ISGID ; } else +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) inode->i_gid =current_fsgid (); +#else + inode->i_gid = current->fsgid; +#endif inode->i_mode = mode; inode->i_ino = objectid;
@@ -4164,11 +4172,13 @@return -EINVAL ; } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )static int btrfs_fiemap (struct inode *inode,struct fiemap_extent_info *fieinfo,__u64 start,__u64 len) {return extent_fiemap (inode, fieinfo, start, len, btrfs_get_extent); } +#endif int btrfs_readpage (struct file *file,struct page *page) {
@@ -4500,7 +4510,11 @@kmem_cache_free (btrfs_inode_cachep,BTRFS_I (inode)); } +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 ) static void init_once(void *foo) +#else +static void init_once(struct kmem_cache *bar, void *foo) +#endif {struct btrfs_inode *ei = (struct btrfs_inode *) foo;
@@ -4523,7 +4537,12 @@ struct kmem_cache *btrfs_cache_create(const char *name, size_t size, unsigned long extra_flags, -void (*ctor)(void *)) +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 27) + void (*ctor)(void *) +#else +void (*ctor)(struct kmem_cache *,void *) +#endif +) {return kmem_cache_create (name, size,0 , (SLAB_RECLAIM_ACCOUNT |SLAB_MEM_SPREAD | extra_flags), ctor);
@@ -4942,7 +4961,12 @@return __set_page_dirty_nobuffers (page); } +#if LINUX_VERSION_CODE >KERNEL_VERSION (2 ,6 ,26 ) static int btrfs_permission(struct inode *inode, int mask) +#else +static int btrfs_permission (struct inode *inode,int mask, + struct nameidata *nd) +#endif {if (btrfs_test_flag (inode,READONLY ) && (mask &MAY_WRITE ))return -EACCES ;
@@ -5036,7 +5060,9 @@ .removexattr = btrfs_removexattr, .permission = btrfs_permission, .fallocate = btrfs_fallocate, +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 ) .fiemap = btrfs_fiemap, +#endif };static struct inode_operations btrfs_special_inode_operations = { .getattr = btrfs_getattr,
ioctl.c +4 -0
@@ -240,7 +240,11 @@return -EEXIST ;if (IS_DEADDIR (dir))return -ENOENT ; +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,27 )return inode_permission (dir,MAY_WRITE |MAY_EXEC ); +#else +return permission (dir,MAY_WRITE |MAY_EXEC ,NULL ); +#endif }/*
modules.order +1 -0
@@ -0,0 +1 @@ +kernel//root/btrfs/btrfs/btrfs.ko
super.c +33 -0
@@ -38,6 +38,7 @@#include <linux/namei.h> #include <linux/miscdevice.h> #include <linux/magic.h> +#include <linux/version.h> #include "compat.h" #include "ctree.h" #include "disk-io.h"
@@ -51,6 +52,9 @@#include "export.h" #include "compression.h" +#ifndef BTRFS_SUPER_MAGIC +#define BTRFS_SUPER_MAGIC 0x9123683E +#endif static struct super_operations btrfs_super_ops;
@@ -609,6 +613,8 @@return ret; } + +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 )static int btrfs_freeze (struct super_block *sb) {struct btrfs_root *root =btrfs_sb (sb);
@@ -617,6 +623,7 @@return 0 ; } +static int btrfs_unfreeze (struct super_block *sb) {struct btrfs_root *root =btrfs_sb (sb);
@@ -625,6 +632,27 @@return 0 ; } + +#else + + +static void btrfs_write_super_lockfs (struct super_block *sb) +{ +struct btrfs_root *root =btrfs_sb (sb); +mutex_lock (&root->fs_info->transaction_kthread_mutex); +mutex_lock (&root->fs_info->cleaner_mutex); +} + + +static void btrfs_unlockfs (struct super_block *sb) +{ +struct btrfs_root *root =btrfs_sb (sb); +mutex_unlock (&root->fs_info->cleaner_mutex); +mutex_unlock (&root->fs_info->transaction_kthread_mutex); +} +#endif + +static struct super_operations btrfs_super_ops = { .delete_inode = btrfs_delete_inode, .put_super = btrfs_put_super,
@@ -637,8 +665,13 @@ .destroy_inode = btrfs_destroy_inode, .statfs = btrfs_statfs, .remount_fs = btrfs_remount, +#if LINUX_VERSION_CODE >=KERNEL_VERSION (2 ,6 ,29 ) .freeze_fs = btrfs_freeze, .unfreeze_fs = btrfs_unfreeze, +#else + .write_super_lockfs = btrfs_write_super_lockfs, + .unlockfs = btrfs_unlockfs, +#endif };static const struct file_operations btrfs_ctl_fops = {
Quoted 1 line · Lee Trager
Sorry it took so long to reply its been a crazy week for me.
You're lucky it was only a week, I often have crazy months!
Quoted 1 line
I can't believe I made such a stupid mistake.
Back when I started in computers, I thought 16 bits would be more than I would ever need to count the number of times I said that exact thing to myself.
Just another stupid mistake I made... the count overflowed :)
jim
Quoted 8 lines · Chris Mason
Hello everyone,
I've rebased the experimental branch again with my latest performance fixes.
I took out the delayed unlink code, it wasn't making a big enough difference in any benchmarks to justify the complexity.
Quoted 9 lines
I changed the delayed backref code to do delayed processing for all extents. In general it is much faster and uses less stack.
I pulled Josef's enospc work into the master branch and asked Linus to pull it.
I'm going to hammer on the experimental branch for a few days and ask Steve to give it another run.
Done. Results for RAID are updated in history tree http://btrfs.boxacle.net/repository/raid/history/History.html
This gives back a few of the performance improvements made on the tree from the 24th (mail server and random write).
Also, I have reordered the bars with btrfs at the end for 2 reasons. First, makes it a little easier to find the latest btrfs results. Also this fixes a problem with my scripts where I was not picking up the random write odirect tests.
Single disk tests are running.
Steve
Quoted 3 lines
-chris
Quoted 15 lines · Steven Pratt
Quoted 14 lines · Chris Mason
Hello everyone,
I've rebased the experimental branch again with my latest performance fixes.
I took out the delayed unlink code, it wasn't making a big enough difference in any benchmarks to justify the complexity.
I changed the delayed backref code to do delayed processing for all extents. In general it is much faster and uses less stack.
I pulled Josef's enospc work into the master branch and asked Linus to pull it.
Quoted 9 lines
Quoted 3 lines
I'm going to hammer on the experimental branch for a few days and ask Steve to give it another run.
Done. Results for RAID are updated in history tree
http://btrfs.boxacle.net/repository/raid/history/History.htmlThis gives back a few of the performance improvements made on the tree from the 24th (mail server and random write).
Thanks for doing this. I'm a little confused by the output though, somehow our configs are giving opposite results ;)
When I run the rand-write workload with 128 threads here, btrfs gets 6236 ops/sec, and ext4 only gets 1509 ops/sec.
My box only has 5 drives, so there is probably a difference in btrfs' ability to keep all the drives in the array busy.
So, I'll do some more experiments here.
-chris